Gentoo in Crisis, Robbins Offers Solution

mrbadbar writes "Gentoo Linux founder Daniel Robbins says Gentoo's leadership is in crisis. 'the Gentoo Foundation's charter has been revoked for several weeks, which means that as of this moment the Gentoo Foundation no longer exists.' Robbins offers a solution: his return as President of the Gentoo Foundation. According to Robbins: 'If I return as President, I will preserve the not-for-profit aspect of Gentoo. Beyond this, you can expect everything to be very, very different than how things are today.'"

Comments

The solution:

[Anonymous Coward]

Disallow future nigger-use of Gentoo. Do not accept nigger contributions, in the unlikely event that they manifest themselves.

Re:The solution:

[X0563511]

Something needs to be done about all this racist bullshit trolling. This is getting ridiculous.

Re:The solution:

[mrsteveman1]

no no, it'll be ok. slashcode just needs some USE flags.

Re:The solution:

[Anonymous Coward]

Don't worry about him, he's just mad because one day he wanted to get another piece of ass from his son but instead he walked into his room to see the boy gobbling a big black knob.

Oh, and when his daughter gave birth and it was a mixed child instead of being another proud inbred spawn of his incestuous advances. Yeah, fuck them niggers for getting all the sweet redneck-family ass that HE wanted to get.

Re:The solution:

[Zonk (troll)]

Something needs to be done about all this racist bullshit trolling. This is getting ridiculous. Slashdot has done something: moderation. If you don't want to see it, set your threshold up to not see -1 or 0 posts. These posts are either by AC's (starting at 0) or known troll accounts (starting at -1).

I have mine set to show everything but have -1 and 0 collapsed so I can still see good posts by ACs and read the occasionally funny trolls.

Re:The solution:

[X0563511]

Moderation isn't enough. These jerks need to be flayed.

Re:The solution:

[Lord Ender]

I don't really think it's racism. I mean, race has nothing to do with the topic at hand. My suspicion is that these sorts of trolls only use that word because it is the most taboo of all words. It's one of the only words that can really get a reaction out of people, that can end careers. It is a word so powerful, that if it is used while committing a crime, the crime becomes many times more serious. That word is used because we, as a society, give it a special power.

Re:The solution:

[Ilgaz]

Something needs to be done about all this racist bullshit trolling. This is getting ridiculous. I am not in mood to read -1 comments but you can report very serious stuff to Slashdot admins rather than waiting for them to get -1.

Huh?

[jd]

The emerge of the upgraded management package failed? Did you remember to set the right USE flags?

Re:Huh?

[Solra Bizna]

Why were you compiling with MAKEOPTS="-j32768"? What did you really expect to happen?

-:sigma.SB

Re:Huh?

[sinclair44]

-j32768 is indeed quite pointless. I don't specify that, just -fomit-instructions... my Gentoo install take up almost no disk space since I've turned it on! However, apparently it tends to mess up your bootloader; I've been unable to start Gentoo and simply haven't had the time yet to look into it.

Re:Huh?

[Anonymous Coward]

Haha, cue the stupid CFLAGS and USE jokes but Gentoo is not only about using l33t flags to a lot of users. There are some of the reasons I can name, which apply mainly to myself. BTW I'm not saying other source distributions don't have these advantages, it would be interesting to hear about other source distributions which have these features...

- An important point is for security, there is no way you can verify the validity of a binary except if you disassemble and inspect it manually. you can't compare it easily because it's platform/cpu/compiler specific output. Sources on the other hand can be directly compared against the trusted source. So for a lot of security-oriented companies, Gentoo, or other sources distributions, are just a lot more secure than pulling binaries from who knows where.

- USE flags, though ridiculed a lot, are a convenient way to achieve efficiency, if for example I don't need OSS support I just disable that flag. Disabling functinality in this way for a system does make sense since it can avoid you having to install things you don't really need. But I do agree they have gone a bit overboard with them, also they are an unstructured mess.

- as we get more cores compile time will not matter anymore, so that argument against source distributions is also lame

- It's just fun to use because you can tweak a lot and the community is very helpful and tries to help users of all knowledge levels.

Gentoo does have some gripes for me, but they are mainly about the package management system

- the package database is divided into +100000 files, even rsyncing this is slow and places a lot of load on the rsync servers
- it's not a scalable approach, as the number of available packages increases exponentially the system gets problems. A scalable system would avoid having a central database
- the use of python makes everything slow since reading this database and emerging a package is a very I/O bound task

So, I hope the Gentoo will survive since it's kind of unique (to all I know) in it's approach and hopefully the main architect can help for this. BTW I'm not directly into the world of Gentoo politics, just a user for a long time.

Re:Huh?

[DaedalusHKX]

Actually, after they broke the stage1 install, I walked off from gentoo, but until then, I ran a local repository alongside my debian and bsd ones.

Most of my "work" is now done off of a Linux From Scratch setup (though with less patches than their actual versions). Quite useful... even if it isn't an official distro anymore (my version of LFS that is.)

Re:Huh?

[jd]

My gripes with Gentoo are the same as my gripes with every distro that uses some form of package manager. There are too many corner cases (same package, different permutation of mutually exclusive options) that are not adequately handled, there are too many specific compiler/linker flags which must (or must not) be used with specific packages only, there are too many ways of crashing the package managers out there beyond recovery, there are too many namespace collisions from independent packages, and packages aren't being maintained in sync resulting in post-install dependency violations.

Some of these problems can be fixed in any distribution - if enough time and care is put in. Some can only be fixed in source-based distros like Gentoo. Some can't be fixed at all without the use of herustics and a lot of compute time. The only way I can think of to produce a totally generic solution is to use software that already exists for calculating optimal settings and for profiling packages, and produce as comprehensive a table as possible of what permutations work for what systems.

Re:Huh?

[PastaLover]

I'll have a look at your "arguments" why don't I.

- An important point is for security, there is no way you can verify the validity of a binary except if you disassemble and inspect it manually. you can't compare it easily because it's platform/cpu/compiler specific output. Sources on the other hand can be directly compared against the trusted source. So for a lot of security-oriented companies, Gentoo, or other sources distributions, are just a lot more secure than pulling binaries from who knows where. Name one company that does this. You do realise every single package they install will have gentoo specific changes which they need to vet. It'll take a while I'm sure. Also, if you were crazy/rich enough to do this, you must as well just download the source packages for any other distro.

- USE flags, though ridiculed a lot, are a convenient way to achieve efficiency, if for example I don't need OSS support I just disable that flag. Disabling functinality in this way for a system does make sense since it can avoid you having to install things you don't really need. Most certainly you can win on one "efficiency" metric and that is installed package size. It's also the one most people don't find very important and everyone else (in the embedded space) is already using custom distros for. Really, USE flags aren't very useful unless in corner cases.

- as we get more cores compile time will not matter anymore, so that argument against source distributions is also lame It'll still take longer than just unpacking the binary. Not to mention all the stuff you'll need to configure. So that argument still stands.

- It's just fun to use because you can tweak a lot and the community is very helpful and tries to help users of all knowledge levels. The only valid argument you've given so far. Yes, gentoo is fun to play around with (for some people). It's like people who tune their car engines, it's a fun hobby. But I still don't see a convincing reason to use gentoo in any other capacity, which is why it has been ridiculed in the past.

One last point,

- the use of python makes everything slow since reading this database and emerging a package is a very I/O bound task AFAIK python makes CPU bound tasks slower, I/O shouldn't suffer much. Just to clear that up.

RIP The Gentoo Foundation

[Anonymous Coward]

you were powerless niggers in a world full of white men

Easy solution

[Anonymous Coward]

$ emerge leadership

Re:Easy solution

[Jessta]

$sudo emerge -av leadership
password:

These are the packages that would be merged, in order:

Calculating dependencies... done!
[ebuild U ] vitural/leadership-3.0_rc2 [1.0_rc1] USE="developers minimal intelligent paludis -emerge -designers " 50 kB

Total: 1 package (1 upgrade), Size of downloads: 50 kB

Would you like to merge these packages? [Yes/No]
Y

Re:Easy solution

[derago]

gentoo # emerge leadership
Calculating world dependencies ...done!

!!! Error: the dev-libs/leadership package conflicts with another package.
!!! both can't be installed on the same system together.
!!! Please use 'emerge --pretend' to determine blockers. ...

[blocks B ] dev-libs/trustees (is blocking dev-libs/leadership-1.3.2)
[blocks B ] dev-libs/nocharter (is blocking dev-libs/leadership-1.3.2)

Re:Easy solution

[arth1]

Calculating dependencies... done!
[ebuild U ] vitural/leadership-3.0_rc2 [1.0_rc1] USE="developers minimal intelligent paludis -emerge -designers " 50 kB

You're touching one of the core problems there. Which needs the following solution:

mkdir -p /etc/portage/profile && \
touch /etc/portage/profile/use.mask && \
echo paludis >>/etc/portage/profile/use.mask

Honestly, when something is this controversial, like paludis is, it needs to either die for the greater good ("kill your babies" as they say in Hollywood), or a fork happen.

Re:Easy solution

[thr33way]

Yeah, but why fork when you can continue to just leech of Gentoo's infrastructure?

Re:Easy solution

[Anonymous Coward]

Actually:
[ebuild U ] app-admin/leadership-3.0_rc2 [1.0_rc1] USE="soft-skills effective intelligent -paludis" 50 kB
[blocks B ] sys-apps/paludis (from pkg app-admin/leadership-3.0_rc2)

Actually paludis never worked well with any version of leadership - collisions all over the place. Leadership-3.0 takes that into account and thus blocks paludis until it is wellbehaved again.

But you might have one point though: All the collisions caused leadership-1.0 to take so much damage that it was not only virtual, but also abstract: No instances running and you were not even able to instantiate new ones.

Re:Easy solution

[stevey]

Are you sure there isn't a missing virtual/cabal in there somewhere?

I know Debian depends upon one ;)

Re:Easy solution

[Anonymous Coward]

Honestly, when something is this controversial, like paludis is, it needs to either die for the greater good

If you can't tolerate the idea of other people having their own projects, instead of going along with what you think they should be doing, then it is you who needs to die for the greater good.

Re:Easy solution

[arth1]

I tolerate the idea -- thus the word fork in my post.

Re:Easy solution

[Anonymous Coward]

But you still demand that they choose one of your two approved solutions. Who the fuck do you think you are to tell people they can't write a new package manager for an existing distribution?

Re:Easy solution

[arth1]

One of the people contributing money to the Gentoo Foundation. Now who the fsck are you, Aanonymous Coward?

Re:Easy solution

[Anonymous Coward]

One of the people contributing money to the Gentoo Foundation.

And that gives you the right to tell other people (who are not affiliated with the Gentoo Foundation) what they can and can't do? Does that mean that if I give money to the WWF I can tell you to go kill yourself?

Re:Easy solution

[Anonymous Coward]

Hahaha, I just noticed... you're this retard? "No, I expect that devs listen to [the users], and respect their opinion as as much worth as their own." LMAO!!!! Seriously, kill yourself now, you worthless, pathetic, sociopathic, authoritarian control freak sick of shit.

G.E.N.T.O.O.

[Anonymous Coward]

The project to Give Each Nigger Two Orphaned Ogres. Brought to you by Felicia Gygax.

Things will be different how exactly?

[Anonymous Coward]

As far as a nigger's concerned, nothing's changed because as everyone knows, NIGGERS PREFER SLACKWARE.

Robbins has my respect.

[GeneralEmergency]

Not too many folks could pen such an offer with out tossing in the phrase "tail between your legs" somewhere.

Re:Robbins has my respect.

[Anonymous Coward]

I was pretty impressed when he actually helped out a guy who had a colo at our datacenter. Nobody with any fame or cred, just some guy who was having gentoo problems that nobody in the community seemed able or interested in helping him out with. Most of us seem to get burned out of helping even relatives pretty early in the game, so doing support for people on the street out of the goodness of your heart is pretty amazing. Even if it is your distro.

Re:Robbins has my respect.

[Anonymous Coward]

He used to run Linux support for IBM. Still Google-able, his answers got me going at work. Smart dude.

Best of luck to Daniel Robbins & co.

[Anonymous Coward]

Tough news, but I trust in Mr. Robbins' ability to get things finished, and I believe in his talent of organization and nigger-slapping, the latter ability of which he was considered an Olympic-calibre competitor.

Gentoo will survive. The community will see to that.

Thanks for all you've given us, Dan.

What is the crisis?

[Kristoph]

I RTFA but I have no idea what the problem actually is that he is going to solve. Could someone explain?

Re:What is the crisis?

[raptor386]

I believe the issue is that the legal entity no longer exists, so he's going to step up, renew the charter, and get Gentoo Foundation recognized as a legal non-profit organization again. Though I understand that this is the issue, I don't understand WHY it's an issue. Hopefully someone else can clarify further.

Re:What is the crisis?

[Jeff DeMaagd]

It looks to be a bunch of internal politics. There's a lot of information that is missing, like why the charter was revoked. Another question is why this guy thinks that he's going to get this power if he's going to replace the people that need to approve him.

The basic information is apparently on a mailing list, which I don't feel like reading.

Re:What is the crisis?

[jmdc]

TFA refers to a previous blog entry, which mostly explains things. To summarize: the people who are supposed to be in charge have mostly resigned or are MIA. The remaining leadership isn't doing things like updating the website, etc - the weekly newsletter hasn't been published in months. The real crises is that they didn't file routine paperwork with the state, which puts the legal status of the gentoo foundation in jeopardy. No one explained why to the community, or said much of anything. So, he's going to get the legal matters cleared up and find new people to be in charge.

Re:What is the crisis?

[foobsr]

Hopefully someone else can clarify further.

The same blog can.

"I am still upset that the Foundation has not been run properly over the last three years, and that many trustees apparently decided to take extended vacations from the project shortly after becoming a trustee, leaving the work to be done by very few - and often a single individual, which defeats the whole purpose of having multiple trustees to do the work rather than a single leader. I am also, like many of you, not happy at all with the way Gentoo has been going from a development and community perspective."

You might also infer what was wrong by looking at what would be different.

CC.

Re:What is the crisis?

[rjames13]

Why does the Gentoo community need these people? Can't they just sail along fine without them? Make their own websites and stuff if they need it?

What is so important about this Gentoo Foundation? What happens if it goes under?

Re:What is the crisis?

[Curtman]

Hopefully someone else can clarify further.

There's at least one response on Planet Gentoo so far. Maybe it will help.

Re:What is the crisis?

[arth1]

The problem with them being unresponsive and MIA is why I fear this offer will fail too -- he wants an answer within a week, but the procrastinators and AWOL people won't give him an answer within a week -- they haven't managed to answer anything else in a timely matter, so why would they suddenly do so now?

Re:What is the crisis?

[Rich0]

I think that one big problem is that much of the gentoo leadership is technical. If a debate opens up over how some aspect of the project is managed, the usual rallying cry to bring everybody together is for all the project leads to talk about what positive things are going on with various technical aspects of the distro.

Now, that is very good in one sense - since we do need to remember the big picture. However, stuff like having a newsletter and all that isn't entirely unimportant. Not having a functional board of directors is a big problem. However, I've been reading the -dev group for months (and on and off for years) and I had just assumed (probably like many others) that this part of gentoo was just going along fine.

To the 20-year-old coder who just wants to create some nifty installer or bootup routine having a board of directors may seem a bit silly. However, if some domain squatter grabs gentoo.org because it didn't get renewed and you can't sue for it back because you don't have any legal standing in any court worldwide then there is a problem. I think that gentoo just tends not to appeal to the sorts of people who like taking care of this stuff - largely because it emphasizes pragmatism and technical achievement - while other distros like debian have an appeal to the kinds of folks who love to read licenses since they make a big deal about that kind of stuff.

I think that the criticality of this "crisis" is a bit overblown. Yes, its a problem and it really does need to be taken care of - expeditiously. However, the world isn't about to end. I'd probably call for rapid trustee elections to fill slots (I'm sure lots of people with half-decent qualifications would be willing to step up), and then have the trustees take action. Since legally gentoo is in quasi-existence it might be possible to not have as much process around all of that - since you can't violate bylaws that aren't binding and all that. But I'm not a lawyer (and the trustees would do well to talk to one).

Re:What is the crisis?

[gentooligan]

Well, he's not offering to just sort out the legal status of the gentoo foundation. There are some conditions attached as well. Basically he wants card blanche to do with gentoo as he pleases (Note, I'm not saying he doesn't have Gentoo's best interests in mind). That means the elected council charged with day to day management of gentoo is made superfluous (unless he chooses to give it something to do). The current developer base will have no say in anything beyond what Robbins chooses to grant them.

The current problems with weekly newsletters, site updates and releases are not something that can be changed simply by resolving the legal status of the gentoo foundation. That said, there are developments in those areas so some improvement might show up in the not too far future.

Re:What is the crisis?

[ImaLamer]

He's worried that the Gentoo product will move to a pay system, where you've GOT TO FREAKING PAY TO COMPILE SOFTWARE YOURSELF!!

How could the product change, by losing its non-profit status, as the summary says? Does someone think I'd like to pay to do all the work on my machine? Just rename the project Ubuntu, that seems to be pretty popular.

(All this is just a joke, god i need to refill my coffee)

Re:What is the crisis?

[/ASCII]

Not for profit does not imply charging money for the downloads, it refers to not turning the Gentoo foundation into a for profit company, like Mozilla, Red Hat and Ubuntu have all done (or rather, have always been). All those project still ship their software free of charge (and are even willing to pay the shipping costs to send you a CD in the case of Ubuntu), but they all have some form plan to turn a profit, like providing for-pay support or living of add revenues.

Re:What is the crisis?

[wellingj]

Yea really, in this day and age do we actually need a 'leader' or is it not better do let things form up however they may.
I think we know what ESR would propose...

Re:What is the crisis?

[dido]

I think that gentoo just tends not to appeal to the sorts of people who like taking care of this stuff - largely because it emphasizes pragmatism and technical achievement - while other distros like debian have an appeal to the kinds of folks who love to read licenses since they make a big deal about that kind of stuff.

So you're saying that Gentoo has more of the Linus Torvalds mindset, whereas Debian has more of the Richard Stallman mindset? Interesting point. I moved to Gentoo largely because I was fed up with the RPM-based distros I had been stuck with until then. Never really got around to trying Debian, but I may soon make the switch, if these troubles really get out of hand.

Question regarding Gentoo

[Anonymous Coward]

Not to sound like an ingrate but I don't want to keep this distro installed if any nigs were involved in the project. At least with Ubuntu, they tell you right up front with their nigger-speak. Can someone clarify?

good!

[kurt555gs]

I never liked the condescending attitude of those Gentoo users that think compiling everything was always so superior to Yum or apt-get.

They are prolly made up of the survivors of the sinking Slackware bunch ( who are equally annoying ).

Of course Gentoo or Slackware or the like will work fine, but in these days of fast processors and cheap memory, why not just use a Debian based Linux like Ubuntu WITH a GUI, and let some one else compile the thing.

There, I feel better now.

Re:good!

[Mantaar]

It's strange how people think Gentooers are into Gentoo for the '--fomg-optimize' thing...

I had to leave Gentoo a few weeks ago because my Laptop couldn't take the massive compiles anymore - my desks are all FreeBSD btw. What I enjoyed about Gentoo was the ports-like package manager and the ability to carefully choose your dependencies via USE-flags. Here I am, back on Debian, and I think it's actually faster... but I don't really care about speed since I exclusively use XMonad and the console - no need for speed improvements on a 1.6 GHz machine with that.

But what I hate is that I don't have overlays anymore. You could dynamically replace any part of your package repository with something you found on the net. Like the proaudio overlay. Or the Haskell overlay. With Debian, this is much harder, as you have to find someone on the Net that will offer his repo of binaries ... people are much less likely to offer that since writing an ebuild is easy, but compiling that stuff for different archs is actually not that easy.

For example, I still didn't find any place that offers a .deb of the new Firefox Beta 3. Anyone willing to point me to one?

The speed is only a minor advantage of Gentoo and manifests itself in the much shorter start up times and the ability to easily switch to baselayout2 or einit to even improve that one. But since the average uptime of my laptop is about 2-3 weeks, I don't really care if Debian takes 20 seconds longer to boot up.

Re:good!

[segedunum]

I never liked the condescending attitude of those Gentoo users that think compiling everything was always so superior to Yum or apt-get.

If you've ever had to hunt around for a package repository because your distribution does not provide, or no longer provides, updates for particular packages and you have no upgrade path - necessitating downloading the source and compiling yourself or completely upgrading your distribution to the latest and greatest - you'll know why the condescending attitude of binary repository developers that everything should be in a repository, and their derision of using source code as a solution, pisses a lot of people off. On top of this, try multiplying this up for different platforms,

When you have experienced this, come back and comment.

Re:good!

[syzler]

Of course Gentoo or Slackware or the like will work fine, but in these days of fast processors and cheap memory, why not just use a Debian based Linux like Ubuntu WITH a GUI, and let some one else compile the thing.

Slackware is package based, although I will admit that the packages do not perform checks for required packages. I only know one person that re-compiles Slackware by hand, but he is a bit eccentric. Most of us only compile packages when they are not included in the distro, when they are not compiled with the options we want, or when we re-compile the kernel.

  It seems odd to lump Slackware and Gentoo together since most of the people that I know who use Slackware are more server centric than desktop centric and prefer stable software that does not change. Most Gentoo users I know are desktop centric and want the latest greatest untested software. This is reflected in the different methodology of the two distros.

I would also like to point out that Slackware has been around longer than Debian and Slackware produces stable releases faster than one every two to three years. Although Debian is a decent distro, there are other decent distros as well. I could argue that a Torx screw is better than a roofing nail, but it really depends on the job at hand.

Re:good!

[Dice]

Gentoo is subject to the same problem in reverse - except it's far more annoying and time consuming.

This scenario has happened to me multiple times on production systems:

Updates get pushed out, glsa-check notifies you of some critical patch to openssl or whatever, you go to do the upgrade only to discover that the new version has a .so rev bump. Now you need to use revdep-rebuild to track down every package that links against openssl (i.e. anything important) and recompile them. If any of these packages are more than a minor revision or two behind what's currently in portage the only way to rebuild them is to pull the ebuild from /var/db/pkg and copy it into the portage tree manually, then rebuild the digest and hope to god that portage can track down all of the source files or that they're still sitting in /usr/portage/distfiles. In the meantime you'd better hope that you're either on a dev box (luxury!) or nobody sneezes, since everything that needs the package that was so bumped is now running off cached filesystem data.

It's a lot of fun.

Re:good!

[MyDixieWrecked]

As has been discussed before, Gentoo isn't an enterprise production OS... in fact, it's not totally ideal for even a single server in a small shop.

The thing about gentoo is that it gives you super-fine grained control over your packages. You want ldap support? want to not support jpeg, but to support png? do you want the package installed, but omit all the X11 bullshit? Or how about keeping a specific version of a package from upgrading when you upgrade your system? That's the power of gentoo's package management system.

Gentoo also offers insight into the innerworkings of the linux OS. You get to build your own kernel and pick EXACTLY what gets installed.

Since Gentoo is frequently on the bleeding edge, it's great for testing out new versions of applications. One of the downsides of CentOS that I've encountered was the fact that subversion isn't quite up to date, and it took several months before vim7 was in the yum repository. Of course, you could add new repositories to yum, or download an RPM specificly of what you want, but that sometimes involves waiting for someone to make the RPM or finding the repository that has what you need.

Another downside of Gentoo, especially in a production environment, is that since it's bleeding edge, many things in the system are changing and usually with a frequency that defies belief. I've been running Gentoo on my own two personal servers (hosting my websites and mysql and DNS and stuff) for nearly 5 years. The sheer number of times that I've booted the machine after doing an 'emerge -u world' and gotten "this configuration file's syntax is depricated, please use this new syntax instead" messages has been infuriating. Routine upgrades aren't routine. You can spend hours picking through config files and manually inspecting the diffs between versions. You don't want Gentoo on your server unless you enjoy spending a day doing an upgrade.

Gentoo is ideal for embedded projects and systems that aren't going to change. The OS lends itself well to projects such as DVRs and controller OSs for robotics. It's small and runs on a lot of different hardwares.

I'm always amazed at how much hate people have for gentoo because you have to build it yourself, but you don't hear people getting mad about the .tar.gz source files they download from sourceforge. You don't hear people bitching about Linux from scratch. The nice thing about Gentoo over LSF is that it automates a lot of the process for you and allows you to set up your system by itself, without the aide of another machine to get the system bootstrapped and initially configured. Sure, some gentoo users are cocky; but they're cocky in the same way that a guy who built his own Camarro acts around their buddy who just bought his new, shiny Saturn.

Gentoo is an exercise in academia. For a user new to Linux who wants to get a feel for the ins and outs and get used to the commandline really fast, gentoo is for them.

Re:good!

[grahammm]

The alternative in that situation is to 'take the plunge' and upgrade all the dependent packages to the latest (presumably stable, as if you are running ~arch then they would likely not be behind) version in portage. As you are talking about production systems, it makes sense to have testing systems which are kept (reasonably) up-to-date so that you do not get (many) unpleasant surprises when updating the live production systems.

Re:good!

[Jack Malmostoso]

For example, I still didn't find any place that offers a .deb of the new Firefox Beta 3. Anyone willing to point me to one?

The best thing is that it's right on your computer, just a couple of commands away:

$ wget http://releases.mozilla.org/pub/mozilla.org/firefox/releases/3.0b2/linux-i686/en-US/firefox-3.0b2.tar.bz2
$ mkdir -p debian/DEBIAN
$ mkdir -p debian/opt
$ tar -xjvf firefox-3.0b2.tar.bz2 -c debian/opt
$ mv debian/opt/firefox debian/opt/firefox3
$ apt-cache show iceweasel > debian/DEBIAN/control
$ joe debian/DEBIAN/control
$ dpkg-deb --build debian
$ mv debian.deb firefox3_3.0+b2_i386.deb

Remember to modify the debian/DEBIAN/control file to look like it makes sense, pretty much like this

Package: firefox3
Priority: optional
Section: web
Maintainer: Yourname <Your@email.address>
Architecture: i386
Version: 3.0+b2
Depends: debianutils (>= 1.16), fontconfig, libatk1.0-0 (>= 1.20.0), libc6 (>= 2.7-1), libcairo2 (>= 1.4.0), libfontconfig1 (>= 2.4.0), libfreetype6 (>= 2.3.5), libglib2.0-0 (>= 2.14.0), libgtk2.0-0 (>= 2.12.0), libhunspell-1.1-0 (>= 1.1.6-1), libjpeg62, libnspr4-0d (>= 1.8.0.10), libnss3-0d (>= 3.11.7), libpango1.0-0 (>= 1.18.3), libpng12-0 (>= 1.2.13-4), libstdc++6 (>= 4.2.1), libx11-6, libxft2 (>> 2.1.1), libxinerama1, libxp6, libxrender1, libxt6, procps, psmisc, zlib1g (>= 1:1.2.3.3.dfsg-1)
Suggests: iceweasel-gnome-support (= 2.0.0.11-1), latex-xft-fonts, libkrb53, mozplugger, xprint
Conflicts: firefox (<< 2.0+dfsg-1), mozilla-firefox (<< 1.5.dfsg-1)
Description: lightweight web browser based on Mozilla
Iceweasel is a redesign of the Mozilla browser component, similar to
Galeon, K-Meleon and Camino, but written using the XUL user interface
language and designed to be lightweight and cross-platform.
.
This browser is based on the Firefox source-code, with minor
modifications. Historically, this browser was previously known as
Firebird and Phoenix.
.
This package is built from the binaries downloaded here:
http://releases.mozilla.org/pub/mozilla.org/firefox/releases/3.0b2/linux-i686/en-US/firefox-3.0b2.tar.bz2

While it's not like downloading a .deb and installing it, it sure is damn faster than recompiling firefox.

Oh and yes, you could just untar firefox in /opt and make a symlink in /usr/local/bin, but you wanted a .deb.

Re:good!

[borked]

Gentoo isn't an enterprise production OS... in fact, it's not totally ideal for even a single server in a small shop.

I'm sorry, but that is total crap. I have been using Gentoo on production servers which I *do* keep current using stable (not bleeding-edge) packages. This is a large shop with many servers. I have never looked back since switching to Gentoo. Everyone who moans about emerges failing and having to run revdep-rebuild often must be doing something wrong. I've had to run revdep-rebuild once when I upgraded libexpat. So what? It took like 2 minutes.

Don't make sweeping statements if you don't know what you are doing. I run Gentoo on my servers and I run Gentoo on my personal desktop and and laptop and have *NO* problems with it. The next time you feel like bashing it, try it first and this time RTFM. Sheesh....

Re:good!

[Anonymous Coward]

Holy Moley!
Gentoo users are cocky in the same way that a guy who built his pedal powered Yugo acts around a buddy who bought a brand new Mercedes. Like people who ride hand built recumbent bikes or use mouseless user interfaces but also make a point of pointing out to everyone around them how that little piece of self-inflicted martyrdom makes them better than everybody else in a way not completely apparent to mere mortals.

Want to learn the command line? Get Unix (any flavour that floats yer boat)

There is nothing wrong with using Gentoo.. Just dont become a Gentoo User.

Re:good!

[Anonymous Coward]

So dumb an answer it is, young padawan.
Learned from Debian then Gentoo, tried FreeBSD, used Solaris 10 for a full year, having a MacBook Pro running Mac OS X as my laptop => my server and my Desktop PC are running Gentoo.
If you RTFM, less problems than Ubuntu or even Debian. FreeBSD is also good but not for a Desktop.

Re:good!

[Anonymous Coward]

Oh and yes, you could just untar firefox in /opt and make a symlink in /usr/local/bin, but you wanted a .deb.

I'm glad you say that, because this is not a .deb. It's just a tarball with a description file and an alternative extension, just like those generated by CheckInstall.

Re:good!

[ttldkns]

The sheer number of times that I've booted the machine after doing an 'emerge -u world' and gotten "this configuration file's syntax is depricated, please use this new syntax instead" messages has been infuriating. Routine upgrades aren't routine. You can spend hours picking through config files and manually inspecting the diffs between versions. You don't want Gentoo on your server unless you enjoy spending a day doing an upgrade.

On my server most of the time i find that etc-update takes care of most of the config file updates. If you have config files you want to protect from minor updates use CONFIG_PROTECT in make.conf to protect specific files. Then the majority of config updates flagged with etc-update are ones which you don't need to read over like tweaks to init scripts or such like. I find my config updates are usually over in a few minutes.

The thing i don't like about gentoo is that after a few years of repeatedly upgrading my system using emerge and building new kernels around new hardware and stuff i have started to feel really isolated from any sense of community or identity gentoo may have. I'm not the kinda guy who hangs out on IRC all the time but with other distros i've used in the past i've really had a sense of direction of where the distro is headed in the future and the grand goal of the project. Gentoo just seems to be like that lazy teenager whos just bumming away his life with no plans for the future.

overall though i think its one of the best distros i've used from a low resources server perspective. It still works after 4 without having to scrap it and start again so i'm not going to switch no matter how lethargic its attitude may feel.

Re:good!

[penguin phil]

There's a Firefox 3 repo for Ubuntu here.

Re:good!

[Anonymous Coward]

I have, I built it from source, it was easy. Checkinstall even automatically generated a package for me.

Try again fuck face. The only reason you're pissed off is because you're becoming irrelevant, as the abandonment by your own foundation clearly demonstrates.

Re:good!

[rbanffy]

"For a user new to Linux who wants to get a feel for the ins and outs and get used to the commandline really fast, gentoo is for them"

That's a good way to ensure the newbie keeps badmouthing Linux, pointing out the chores required just to keep it working, in front of his fellow Windows users.

A newbie needs a gentle introduction to Linux first. Ubuntu (disclaimer: I use it) is just fine - it's good enough, pretty and mostly works. In fact, many of my gurus prefer Ubuntu or Debian manly because it lets them focus on what they are doing when whatever they are doing is not maintaining their boxes. If the newbie then feels like he wants to go deeper, then, maybe, Gentoo would be interesting. Or perhaps, it would be educational, if not too scary, to build a Linux box from scratch.

Gentoo is the best choice for the hackers who love to manage their computers.

Re:good!

[Anonymous Coward]

Yeah but you need to really think about how much time you spend maintaining it. Seriously think about it. Even if it hasn't cost you much time, it probably will at some point. This is what people tend to realize about Gentoo after using it for a while. It's fun to hack and at first but after you have spent enough time with it and used some of the alternatives you start to realize that it just isn't worth the trouble when you need to get real work done.

Re:good!

[Jack Malmostoso]

Slightly better since I copied the dependencies from the official iceweasel package, but I agree with you.
A package built like that should *not* be redistributed, just like those created with checkinstall.

Re:good!

[garutnivore]

MyDixieWrecked, you present a reasonable picture of what Gentoo is good at. Thank you for that. The problem is that too many Gentoo advocates out there are asses about the choice they made. Here's an all too frequent scenario. You're discussing a problem in Ubuntu or Fedora, trying to help someone get the system running correctly. Then, a Gentoo advocate comes by, screams "Ubuntu (or Fedora) is teh suxx0rs! Gentoo, FTW" and then tries to convince every one and their pets with juvenile arguments that Gentoo is the best distribution out there, irrespective of what the end user's needs may be. (Yeah, I'll put my mother on Gentoo.)

Most likely, such Gentoo advocates form a minority of Gentoo users but they form a vocal minority. The problem with such vocal minorities is that they often are so vocal that people start thinking that these minorities represent the view of the majority. I think a fair amount of the "hate people have for gentoo" comes from interaction with those jackasses.

Someone might ask what about Fedora users acting like jackasses or Ubuntu users, or Slackware users. Well, those exist too but somehow Gentoo jackasses seem more frequent to me.

Re:good!

[Anonymous Coward]

Another downside of Gentoo, especially in a production environment, is that since it's bleeding edge, many things in the system are changing and usually with a frequency that defies belief. That isn't a downside. Quite the opposite. One example is CentOS 5 which came with a version of PHP that didn't have the ability to display the progress of a file upload to the user. Even if Gentoo didn't have an ebuild in their repository for that version of PHP (and no one had posted one on Gentoo Buzilla), it'd be trivially easy to copy the exiting ebuild script into your local portage overlay (/usr/local/portage), bump a version number and update to the newest version of PHP which has the feature you need. And if you were worried about other changes in the new version causing problems, you could just create a diff of the required feature in the new version and backport it to the old version you're using with about 2 additional lines in the ebuild script. Of course, this is more useful for security patches and small bugfixes.

You don't have to upgrade to the latest version of a package if you don't want to. You can keep using 2 year old software on your Gentoo server with 600 days uptime and manually add security patches to software as required (just a diff + few extra lines in the ebuild you're using). Binary distributions do the backporting for you but because the packages are all built generically with every feature enabled, you'll find that you have to update the package more often than if you manually did it via the "Gentoo method".

Most Gentoo users would just go for the approach of using the latest packages but they'd first test them out on a development server. A real production environment with a focus on zero downtime would have multiple geolocated servers working together for the same cause (example: round-robin DNS, distributed MySQL databases, multiple Apache/Squid frontends, etc). All changes would be tested before pushing them out to the real world. When people say that a Gentoo system can't be used in mission critical enterprise environments, they're speaking complete nonsense. Downtime has little to do with the distribution and almost everything to do with the system administrators and the network/system architecture.

The sheer number of times that I've booted the machine after doing an 'emerge -u world' and gotten "this configuration file's syntax is depricated, please use this new syntax instead" messages has been infuriating. The fact that you're constantly rebooting your servers (without checking for configuration changes after updates) indicates you don't know much about running servers. Are you sure you've used Gentoo before? I am taking a guess that you're complaining about the change between Apache 1 and Apache 2 configuration formats (or something on a similar scale)? There is NO part of the Gentoo packaging system that would complain about deprecated configuration file formats on a reboot (but individual software packages would complain). Emerge will give you courtesy warnings about configuration changes after the emerge (update) has completed. Just run "dispatch-conf" to either automatically merge configuration files (if you haven't modified them since the last installation of the package) or do a simple diff-merge between the old and new formats for minor revision bumps.

A lot of people using binary distributions (sometimes even those who can write MPEG4 algorithms in multiple assembly languages while blindfolded) don't understand that Gentoo is versionless. People often complain about Gentoo pushing large updates like Apache 1 to Apache 2... something that is only possible with binary distributions if you do a huge all-at-once update on a scale more grand than RHEL4 to RHEL5. It makes a lot of sense to use a versionless distribution rather than be faced with having to update 400 packages all at once.

Re:good!

[MyDixieWrecked]

I've had to run revdep-rebuild once or twice; because of expat and one other package, but in the last 5 years, gentoo has changed their config syntax for networks and the way that pam works with ldap logins. I've had to do big jumps in kernel upgrades because of lack of support of features I use in iptables which caused me to have to upgrade to udev from devfs, and I've had several other weekend's worth jobs because of config file and package deprication.

The reason that I say that Gentoo isn't ideal for production is because of these changes that are made. If you want to stay relatively up to date in your packages (for security reasons), it can be a pretty big pain to keep up with config changes. The measures that you can implement to protect your config files aren't always great since sometimes a minor upgrade will incorporate some new configuration option (I've seen this in apache, php and ldap on several occasions).

Rolling your own enterprise-level server management application, I think, would be easiest on gentoo because so much of the system is exposed and it is so straight forward, but my point in my original post is that for production, you'd like a reasonably stable implementation of configuration and some way of keeping my old setup until I'm ready to do a full revamp of my machine.

I didn't bash gentoo. I'm a gentoo user and I like it a lot... I just wouldn't use it for contract jobs or at my real job (sysadmin of around 130 machines); for that, I'd rather use CentOS.

As an aside... although I do enjoy building my own kernel, it can be time consuming and doing large jumps (I've gone from 2.6.4 to 2.6.12 to 2.6.19) can be a real pain in the ass to remember to properly enable all options. I've got an issue right now where the current kernel will not boot my machine and after dozens of attempts, I haven't been able to solve it. I've got several unsolved posts in the gentoo forums about this... the downside is that I'm running gentoo on PPC hardware, so I don't have the largest pool of people to help troubleshoot.

Re:good!

[jimicus]

I'm glad you mentioned that, as it's bitten me more than once in the past.

Gentoo is great to experiment with, and provided you can keep the system bang up to date (and live with occasional breakage), fine. But many of us aren't prepared to make that sacrifice.

On the plus side, maybe as a result of this, it's produced a very helpful community. Much more so than many other distributions.

Re:good!

[Blackknight]

That's why you don't run Gentoo in production, stick with CentOS or Debian stable.

Re:good!

[I'm just joshin]

Or you use your dev machine to build binary packages of everything you need and then have your production machines install those packages. Think of Gentoo more as a distribution -builder- instead. I have 1 box building for dozens. -J

Re:good!

[Anonymous Coward]

So I take it that upgrading from CentOS 4 -> CentOS 5 every few years on 130 machines (26,000 package upgrades at once) to ensure you have the 'latest' features is OK, whereas doing 130 updates once or twice a week with Gentoo is bad? Does CentOS have an easy way to merge configuration files between CentOS 4 and CentOS 5? And how easy is it to compile a custom kernel for use in CentOS?

Most Gentoo complaints I see are comparisons of non-existent features in a binary distro with existent features in Gentoo. I'm constantly having to argue with common/Windows people that use "Ubuntu hasn't got firewall or antivirus protection" as a valid comparison between Windows and Ubuntu. And now that Linux distributions are becoming more popular, I'm finding that a lot of distribution comparisons are just as invalid, nonsensical and biased.

Re:good!

[WuphonsReach]

Tried it for 2-3 years (Gentoo on servers), gave up and have switched to CentOS/RedHat.

I loved Gentoo as a way to learn about Linux on the server. But the project has been seemingly adrift for a while, so we're making the move away from it. That, and it's a bit too fiddly on servers for our tastes. (It seems like it should be a no-brainer for servers, but ultimately it's not quite stable enough.)

Re:good!

[color]

it is hard to switch back to a binary distro after gentoo, but the parent has a point in that gentoo gets harder and harder to manage. That's why I started looking for alternatives and found Gobolinux, a distro that makes it really easy for mantainers and admins. It is a fresh air in the unix world.
www.gobolinux.org

Re:good!

[Deadguy2322]

Freash air in the Unix world? Yeah. Nobody would notice it over the body odour of the penguin molesters.

Re:good!

[Thyrteen]

I'm with you. I ran 2 gentoo servers that I took down just before I moved to colorado. another firewall is still running at work, and my laptop is fine and dandy running gentoo. The servers did get a little overwhelming, and I guess that's dependant upon the number of services you run too, but even apache at times.. gawd.. I remember when they split the main apache config to several files.. try diffing that! The laptop is great, I can get all the cutting-edge wireless tools working no problem.

Re:good!

[Penguin's Advocate]

Amen to that! We run our servers on gentoo as well

Re:good!

[NeedAName777]

If config changes are difficult to someone, mabey they are not using the correct tools.(dispatch-conf) I run an enterprise-level server and I love that Gentoo doesn't arbitrarily overwrite my configs. I really think that Gentoo correctly notifies you the admin that technology has changed (and is deprecated) rather than just going on and letting you think everything is peachy. If someone is using apache,php or ldap they SHOULD be editing these files and verifying there are no unauthorized/incorrect changes as well as keeping up to date with current features. As for kernel config I believe a minimalistic kernel reduces the chance for exploitation/system hangs.

Re:good!

[LinuxDon]

Quote: "As an aside... although I do enjoy building my own kernel, it can be time consuming and doing large jumps (I've gone from 2.6.4 to 2.6.12 to 2.6.19) can be a real pain in the ass to remember to properly enable all options."

Why don't you just copy the .config file over to the new kernel?
Kernel migrations aren't a big issue at all imho.

Re:good!

[CAIMLAS]

Re: a debian package for Firefox beta 3...

Easiest way to go about that is download the tarball, then run alien -d ff3beta2.tgz (or whatever it was called, i forget), then dpkg -i ff3beta2.deb --instdir /opt ; that should do it.

I think you could also extract the tarball, run dpkg -b dirname/ packagename.deb and get similar results.

Now, if you want it to be "compliant" to the debian package standards, that might take a bit more time. But for my purposes, I just want a dpkg db entry so I can remote it when I get a new version and be sure I got the whole thing.

Re:good!

[segedunum]

I have, I built it from source, it was easy. Checkinstall even automatically generated a package for me.

You missed the point dipshit. You're having to handle source here because of the limitations of a binary packaging system. If you're having to handle the source then you're already in a whole different territory, and good luck satisfying the surrounding dependencies.

Try again fuck face. The only reason you're pissed off is because you're becoming irrelevant, as the abandonment by your own foundation clearly demonstrates.

I say something against binary packaging systems and someone immediately thinks I'm some sort of rabid Gentoo supporter (I don't do much with Gentoo today, hence the post about how the alternative is a bit crap). It kind of throws a new light on all that Gentoo ricer stuff people like to throw about as a laugh.

Re:good!

[ion.simon.c]

You must be new here? I've been running Gentoo since 2002. (I ran screaming from RH9.)
I can tell you that I've had to make multiple runs of revdep-rebuild on more than one occasion. (However, all those occasions were within the last year or so... The current package maintainers are fond of tossing out upgrades to core libs that aren't binary compatible with older versions.)

Re:good!

[MyDixieWrecked]

Why don't you just copy the .config file over to the new kernel?
Kernel migrations aren't a big issue at all imho.

I've tried doing that in the past but have wound up with errors. I believe it's because my servers are both PPC and it seems that features appear and disappear and get renamed between versions, at least between gaps of 5+ versions. Going from 2.6.12 to the latest will not work at all. there are an enormous amount of new features that require enabling.

Re:good!

[WWWWolf]

For example, I still didn't find any place that offers a .deb of the new Firefox Beta 3. Anyone willing to point me to one?

Know what I'd do?

Keep current iceweasel packages to satisfy dependencies. Download the official Firefox Linux beta binaries. Stick in... some location. Update your desktop launchers/menus (create a ~/.menu/local.firefoxbeta and do update-menus if you're feeling fancy, or just create a new launcher on desktop or something). Done. Just be careful with iceweasels.

Don't know how FF3b works right now, but I used FF2 this way when it was released and it took some sweet darn time for the .debs to materialise. For me there were no ill effects when going from and back to Debian packages.

I know what you're thinking and I agree: It's not pretty, it's not .deb. But my point is simple: If there's no .deb, it's not that critical. There's many, many ways to handle software packages cleanly. In this case, using the binary gets the job done until Firefox 3 is out for real because the Mozilla folks provide a nice clean binary tarball you can uninstall effortlessly.

A clean directory of binaries is a clean directory of binaries where the heck it happens to be.

This is my general battle plan:

1. If it's on official Debian repository, use it.
2. (Optomatomototize with -fomit-sanity? apt-get source. Not that I'd be into it.)
3. If someone has an APT repository, use it.
4. Uh... random .deb download location? Not good, but better than nothing. Use it.
5. Pre-built binaries? Use them, with care; stick in /usr/local or /opt or ~/Applications or whatever.
6. Source only? Can do. Just use Stow to manage the stuff: --prefix the thing to /usr/local/stow/whatever-1.0.0, make and make install, then do "cd /usr/local/stow; stow whatever".

Stow in particular is particularly cool. Keeps /usr/local just as nice and clean as dpkg keeps /usr =) Uninstalling stuff built from source is as easy as cd /usr/local/stow; stow -D whatever-1.0.0 && rm -rf whatever-1.0.0 ...

Re:good!

[Dice]

If you copy the .config then run "make oldconfig" you'll pull in all of the still-relevant settings from your last build and be asked about anything that has changed.

Re:good!

[Anonymous Coward]

revdep-rebuild once when I upgraded libexpat. So what? It took like 2 minutes.

That was not my experience.

Re:good!

[Mantaar]

Thanks very much for the replies, guys, but unfortunately my initial request was badly worded.

I didn't really want a .deb that much as I wanted a repo. So new versions of FF beta that crop out (which happens every now and then) would then automagically get updated with apt-get update & upgrade.

That's one of the features I like most about Linux (and FreeBSD, for that matter, though there it's a whole different story): not only is the installation of programs incredibly easy, but you can also keep everything up-to-date without having to bother about it yourself. Or have the apps phone home regularly to check for updates. Or worse: have the apps perform stealthy updates. :-)

I'll think about making my own repo then. But I'll need some server space...

Re:good!

[MyDixieWrecked]

ahhh. I'll have to check that out. I was unaware.

Thanks. =)

Re:good!

[turbidostato]

"So I take it that upgrading from CentOS 4 -> CentOS 5 every few years on 130 machines (26,000 package upgrades at once) to ensure you have the 'latest' features is OK, whereas doing 130 updates once or twice a week with Gentoo is bad?"

Yes, you are right.

Even if the two approaches took just the same time after the five-year timeframe, what we will call the "CentOS-way" (not that it has any exclusivity with CentOS, it's just to follow your example) has to be considered *vastly* superior on *any* production environment. The ability to have a given time-frame for a job being done is a must on a corporate environment; while it's relatively easy to define a maintenance window of, say, two months each five years where you think about, design, test, announce and deploy a programmed upgrade on the "CentOS-way", it's not only going to be time-sucking but undoable most of the time the "Gentoo-way" (today is "upgrading monday" but the mail server's main hard disk is failing... well let's suspend today's upgrade, just to have much more problems next week since upgrading depency problems tend to grow exponentially), and will deal to bad press (people can expect minor problems the next two weeks after the "big update" each five years and, while mumbling, they will accept them and will forget for the next half decade); now try to get a satisfied user base with a system that is *guaranteed* to have minor problems almost always (and probably not so minor problems from time to time), not even considering unavoidable minor (or even major) changes each time "foobar" goes from 1.2.3 to 1.3.0 (and there will be at least one of such "foobars" almost weekly).

And then, I'd say that just considering context-switching time will deal to much more time on your "Gentoo-way" but that's nothing but cream on top of the cake.

All of this is just plain evident for everyone that has even the slightest exposure to any real production environment, so you telling that makes me highly suspicious you just hadn't such an exposure (i.e.: you are just making your opinions out of your basement's toy-computing experience). Of course, I may be wrong and then I'd be very courious about your "real life" experiencies regarding your "Gentoo-way" for systems administration and how do you avoid the previously described situations.

Your second paragraph is just too off-topic to be even considered.

Re:good!

[Xtifr]

I would also like to point out that Slackware has been around longer than Debian By a couple of months. Both projects started in '93. Of course, since Slackware was basically just a simple fork of the existing SLS system (which in turn was not much more than a bootstrap tool added to the existing Sunsite archives, which is what I used), it was able to make its first release almost immediately, while the far-more ambitious Debian project took quite a bit longer to reach something they were willing to call "1.x". (Although the 0.9x stuff was pretty solid in my experience.)

While your point that different tools serve different purposes is a good one, I've broken too many Slackware systems over the years to have any desire to return to using it for anything that doesn't absolutely demand it. Perhaps if I worked on embedded systems, it might seem like an attractive option, but I tend to think that I'd prefer something more like Gentoo or FreeBSD in a case like that. Simplicity can be a virtue, but sometimes, too much simplicity can actually make things harder. And that's been my experience with Slackware. Heck, if I just wanted simplicity, I could switch to FreeDOS. Now that's a simple system! :)

Re:good!

[UltraMathMan]

Firefox 3 beta is available for download as a .deb at the bottom of the page.
http://packages.ubuntu.com/hardy/web/firefox-3.0

Re:good!

[lems1]

Get more people to use Ubuntu and all vendors (3rd party) will provide packages for it. Look at Skype, LimeWire, etc... The trend is there now...

I'm converting everybody I know to Ubuntu because it's easy and it just works. Whoever comes to me with a virus problem (windows) or any type of problem that requires me to re-install the OS, i simply tell them: it's free for you if I install Ubuntu. And I give them the choice that if in 2 weeks of using it they don't like it, I'll restore their Windows stuff for free as well.

I have not heard a single complaint. The first few days people ask me "how do I install foo" and I point them to the right menus. Simple.

Tourettes?

[killmofasta]

Are you suffering from Tourettes? It is treatable? Can you have typing Tourettes?

Re:Tourettes?

[Anonymous Coward]

Why would anyone want him as a president? What's wrong with Gentoo leadership? Nothing and everything. The idea of council is cool, but at least 50% of devels are plain fools. I'm not glad to say thins, but it is. What gives me the right to say it out loud? I use Gentoo for years now (since 2002 or 2003, can't remember), and I'm a mid level C/C++ programmer, so I can judge. President? Why? Microsoft worker as a president (you do know he left Gentoo so he could work for Microsoft, right?)? Thank you for kick staring such a good distro, but it's too big for you now, Daniel. Go back to Microsoft!

Re:Tourettes?

[nxsty]

Actually you can. Just install the firefox tourettes extension. :) http://fffff.at/tourettes-machine

Re:Tourettes?

[Anonymous Coward]

Line jumping Troll. They should have a -2 mod.

Re:Tourettes?

[Anonymous Coward]

That's possible. piss out my ass. Still it's fuck shit fuck not very likely.

Trouble

[Frekko]

I left gentoo some time ago due to severe problems. Let me sum up the most problematic ones: 1. Package system becoming VERY VERY slow because of the amount db size. 2. No sane way to upgrade properly without doing several rounds of breaking and fixing library dependencies 3. USE flags change all the time and often leave the apps crippled if you don't set it up "just" right (try PHP) 4. No automatic way to uninstall a package and have the system automatically remove the unused ones. 5. Very very slow upgrade cycle for major packages (KDE is a good example)

Re:Trouble

[Frekko]

Let me add a few more:
6. You are forced to update VERY frequently. More than a month and you are CERTAIN to get issues while compiling.
7. Actually getting a usable desktop (with udev, automounting etc.) working is a hell of a lot of work

Re:Trouble

[clayne]

emerge --sync && emerge -uDN world

Iteratively updates any newly synced ebuilds. Pretty much a no brainer and I think when you say "some time ago" you actually mean 2004.

Re:Trouble

[Anonymous Coward]

7. Actually getting a usable desktop (with udev, automounting etc.) working is a hell of a lot of work

But at least all the copy-pasting of commands from the Gentoo forums makes you feel like you're learning a lot about how Linux works, right? ;o)

Re:Trouble

[Noksagt]

I don't use gentoo much anymore, but I did not too long ago.

1. Package system becoming VERY VERY slow because of the amount db size.

Robins claims there was also a bug in a recent portage version that slowed things down quite a bit.

4. No automatic way to uninstall a package and have the system automatically remove the unused ones.

Like:

$ emerge -C [packagename] && emerge --depclean

or do you mean something else?

5. Very very slow upgrade cycle for major packages (KDE is a good example)

Do you mean slow time to see updated ebuilds or that it takes a long time to compile? 3.58 is in the repos & there is a 4.0 overlay. Think there are even cvs ebuilds floating around.

Re:Trouble

[Frekko]

5. Very very slow upgrade cycle for major packages (KDE is a good example)

Do you mean slow time to see updated ebuilds or that it takes a long time to compile? 3.58 is in the repos & there is a 4.0 overlay. Think there are even cvs ebuilds floating around.

Compiling is not the issue. That is very fast on modern computers. Also, the ebuilds often are in portage alright. The problem however is that it sometimes takes AGES for them to be marked stable. And yes, I know you can "fix" that. However, that usually leads to a cascade of masking issues which you again have to maintain etc.

Yeah, I know the --depclean trick. Unfortunately on my gentoo boxes it crashes (and has to be run several times) more often than I have time for.

Re:Trouble

[ozmanjusri]

7. Actually getting a usable desktop (with udev, automounting etc.) working is a hell of a lot of work

1. Download Sabayon.
2. Boot from the LiveCD
3. Run the installer if you like it.
4. ???
5. Profit

Re:Trouble

[misleb]

But at least all the copy-pasting of commands from the Gentoo forums makes you feel like you're learning a lot about how Linux works, right? ;o)

Ha! That's exactly what I thought when I first installed Gentoo from stage1. I was just following the howto, line for line. Great guide, but I can't really say I learned anything. If they're just going to provide step by step instructions to do everything anyway, why not just make the system easier to install/use in the first place?

Re:Trouble

[Teppic_52]

If they're just going to provide step by step instructions to do everything anyway, why not just make the system easier to install/use in the first place?
You have clearly missed the point, it's not made to be 'difficult', that is just the way it is.
Plenty of people find the annoying idiosyncrasies of Gentoo worth the effort, their reasoning for this is their own and probably unique. If you gain no benefit use something else, if it's too hard for you use something else.

The only prerequisite for being able to install Gentoo instead of any other distro is the ability to read IMHO, and the 8 or so hours I spent 4 years ago installing my 1st system was well worth it as it's the same install I post from now.

Re:Trouble

[Anonymous Coward]

If you have nothing better to do for the next 485 hours of your life then by all means install Gentoo and knock yourself out! I see a bunch of asshole developers at work who spend half their time fixing their computers and trying to get Gentoo shit working instead of doing their fucking job. They are being paid to write code for the company (like I'm doing) not fucking around with their stupid Gentoo box all day long.

Re:Trouble

[cinderblock]

This is what I did with my first Gentoo install. At first I had no idea what was going on. People that only follow the instructions never find out what's happening. But if you explore a little and try some different things while installing you figure it all out. I am not a professional IT guy (I host my home server as a hobby and service for friends and family) but I am confident in dealing with Linux (maybe just Gentoo for now) instead of only knowing the ins and outs of Windows (Isn't that MS wants?) My friend wanted to have his own small server and tried to get Debian to work... Dependency failures everywhere never got the system booting off the HD. I got him to try Gentoo and it worked the first time. At first I thought the compiling for speed was sweet and that I could brag about how efficient I was running but then I realized how little that mattered. It's really just the ease of use of Gentoo that makes it so powerful. (My first Linux box I ran from a RedHat CD I bought. What a disaster. Then again, I was still in middle school.)

Re:Trouble

[Anonymous Coward]

Aww, poor baby gentool. Someone sounds like he's a widdle bit upset about the fact that the people behind Gentoo _themselves_ don't even give a shit about it any more.

Nice elitism too there, "oh, you only have to be literate to install Gentoo, all of the other distributions cater to morons and I'm above them obviously." Do everyone a favour and throw that four year install in the trash compactor rather than subject everyone to another four years of your bullshit, gentool.

Re:Trouble

[Anonymous Coward]

Sorry... I fail to understand how you simply copying and pasting is a failure on Gentoo's part for you not learning. If you go to college and copy off someone else's papers constantly, is it the college's fault you didn't get a good education?

Re:Trouble

[vandan]

6. You are forced to update VERY frequently. More than a month and you are CERTAIN to get issues while compiling.

Fucking bullshit dude. What the fuck are you talking about?

7. Actually getting a usable desktop (with udev, automounting etc.) working is a hell of a lot of work

Oh dear! Add the 'hal' and 'dbus' use flags and emerge a recent desktop. I have it working with NO manual configuration on almost 30 computers. What's wrong with you?

Seriously, if you want to use another distro, for whatever reason ( say, like you're not quite bright enough for a 'roll your own' distro ), then fine. But if you want to constructively add to a conversation, it helps to leave the bullshit in your head rather than have it ooze out your mouth or dial-up line.

Re:Trouble

[thr33way]

Okay, I fail to understand this - the developers feel that they aren't ready to be marked stable... so they aren't marked stable... you think they should be simply because.... what? upstream said its the stable release? that makes it less buggy? I don't understand all these people who constantly bitch about how "slow" things are to go "stable" but refuse to use anything but stable.... If a developer doesn't feel the package isn't ready to go stable... it doesn't go stable. Remember there are other configurations out there beside your own. If you want something "stable" go use a distro where everything is identical. They only need to work out the bugs for the exact settings that they specify that you get the priviledge of using.

Re:Trouble

[Anonymous Coward]

I can tell you how I handle a couple of your concerns on my systems.

1. Package system becoming VERY VERY slow because of the amount db size.

Grr. This is the one I really want fixed.

2. No sane way to upgrade properly without doing several rounds of breaking and fixing library dependencies

My standard upgrade method:
emerge --sync
emerge -avDNut system
emerge -avDNut world
emerge -a --depclean
revdep-rebuild -av

If you haven't been using the -DNu (update, check for new USE flags, deep tree search) flags, then there will probably be a lot to fix. Personally, I have no idea why -D isn't the default; it's the safest thing.

3. USE flags change all the time and often leave the apps crippled if you don't set it up "just" right (try PHP)

In the emerge steps, the new flagds are highlighted in yellow if you use -v. So you can stop the emerge and use euse -i flagname to find out what's going on.

Just saying.

4. No automatic way to uninstall a package and have the system automatically remove the unused ones.

That's what emerge -a --depclean is for. It's got the scary warnings because, if you haven't been using -DNu on your emerges, it can break lots of stuff.

5. Very very slow upgrade cycle for major packages (KDE is a good example)

That one depends on community support. The main problem I've seen is things taken forever to move from unstable to stable, because nobody tells them if it's stable on their machines.

6. You are forced to update VERY frequently. More than a month and you are CERTAIN to get issues while compiling.

Short answer: If you aren't willing to upgrade frequently, in order to get the latest and greatest, then Gentoo is probably not the right distro for you.

Long answer: If you want to use Gentoo on servers, you need to have a test server or servers set up for updating and getting the kinks worked out, and have it set to build binary packages. Then do binary installs from the test servers to production when done. I use this arrangement for updating my firewall; the "test" server is an inside machine with the DHCP and (once I finish setting it up) Nagios servers.

And Nagios is being much more of a pain to set up than Gentoo ever was.

7. Actually getting a usable desktop (with udev, automounting etc.) working is a hell of a lot of work

QFT.

Gentoo has its flaws, but I've found that doing the above upgrade pattern generally works with little or no hassle, aside from waiting for things to build.

I, for one, love etc-update. I wish RPM used it. Same for their named-runlevel init script system.

Re:Trouble

[fellip_nectar]

$ emerge -C [packagename] && emerge --depclean && revdep-rebuild

Fixed that for you...

Oh, and pray that revdep-rebuild doesn't pull in Openoffice... AGAIN.

Re:Trouble

[Anonymous Coward]

What you say is very strange. Are you sure you use Gentoo? Because I really do no have this experience.
I use Gentoo in stable and testing on x86 and x86-64: it's an intense living creature (tens of updates per day, more useflags for packages everyday for testing). It's moving fast and it's damn stable.
Really, the software is moving forward, maybe the paperwork is lagging...
Where you are right: rewrite portage in GPL C.

Re:Trouble

[Sancho]

Stability is a pretty important thing. Look at Debian, for example. Tracking stable on Debian means that you'll be using versions of software that are years old, with only security issues causing updates (and even then, the issue is usually patched against the original source, rather than requiring an upgrade that likely introduces new features and possibly new bugs.) Stable means stable, and it should be very explicit.

If you want the newest and greatest, run the the unstable branch.

Re:Trouble

[ultranova]

Stability is a pretty important thing. Look at Debian, for example.

Mind you, Debian isn't without some weird issues either. For example, VNCServer doesn't depend on any font packages, despite not being able to start without them, with the official explanation being that the user could be running a font server somewhere in their network.

Re:Trouble

[cryptoluddite]

That's all true, but hardly severe except for the VERY VERY slow package system. What really kills me is that it's only slow because there are 200k files scattered everywhere. All it really needs is to put these in a .zip (yes, zip) file. An uncompressed zip can still be sent as a delta with rsync, even a compressed one can with a little work on the adding/removing/updating code. But most importantly it can be randomly accessed unlike a .tar.gz file, and it's easy to access in python.

Just making this one change would improve the gentoo experience a LOT and not really make it feel inaccessibly like an actual db would, wouldn't need any 'db update protocol' or loopback filesystems or anything complicated like other suggested fixes.

Re:Trouble

[hasdikarlsam]

Speaking as someone who uses that packages and *does* run a font server on another machine, this is a good thing.

Re:Trouble

[marcansoft]

PLEASE don't recommend Sabayon for HDD installs. Sabayon is to Gentoo what Knoppix is to Debian. Sure, it works nicely as a live CD, but I hope you don't mind never upgrading if you do decide to install it. emerge -uav world on Sabayon tends to break the entire system.

Re:Trouble

[Anonymous Coward]

The package system is now a good bit faster. The only thing I find the least bit slow is the process of synchronizing to the repositories. A cron job pretty much renders that a moot point.

As far as upgrading, I now upgrade less frequently, and rebuild dependencies each time (emerge -uDN && revdep-rebuild). As far as removing unneeded packages, emerge --depclean has always worked for me.

All in all, for me, it has beat the hell out of re-installing other distro's every year or so. My first gentoo installation lasted four years. I had to re-install when I got a new computer.

Re:Trouble

[celle]

"No automatic way to uninstall a package and have the system automatically remove the unused ones. "

Freebsd doesn't do that either from my experience. It just removes the installed port and doesn't get rid of the installed dependencies even if they're not used anywhere else.

Re:Trouble

[Anonymous Coward]

I am leaving (converting machine #3 today) because in some kind of misguided "quality control" crusade they have been removing packages that I rely on, and breaking the hell out of everything else. It has been months since I've installed or upgraded anything, because I can't get much of anything to build without bombing off.

Re:Trouble

[LinuxDon]

In my opinion it's still much better than the mess that will unavoidably be created by having to compile packages yourself like you need to do with other distributions.
I tried ubuntu once and -very- soon discovered that I needed to compile sshd by hand just because I wanted to use openct. Also it's missing quite a few packages that are a hell to track down. Furthermore, adding other unofficial repositories trashed my system because it kept giving me errors when I wanted to do upgrades.

In my experience, Gentoo provides almost any package and compilation option you'll ever need and is almost impossible to break.

Re:Trouble

[Derek Pomery]

This (extreme) approach might help.
http://m8y.org/gentoosync.txt

Re:Trouble

[misleb]

Sorry... I fail to understand how you simply copying and pasting is a failure on Gentoo's part for you not learning. If you go to college and copy off someone else's papers constantly, is it the college's fault you didn't get a good education?

The difference between installing Gentoo by copying/pasting from the howto and copying other people's work in college is that in the case of Gentoo, that is the accepted way of doing it. Really, it is the ONLY way to do it as far as I can tell. I mean, there's nothing obvious or intuitive about the steps. Even being a seasoned Linux user going into it, I couldn't have guessed the exact commands required to get it up and running.

It isn't Gentoo's "fault' that I didn't learn anything from it. I'm just saying that Gentoo elitists should give up the illusion that the install process is some profound leaning experience/initiation that every Linux user should have to go through.

-matthew

Re:Trouble

[HAWAT.THUFIR]

On the e-mail list there's a particular user who repeatedly points out this defect -- he's shouted down with the adage "gentoo isn't supposed to be easy". Then, there are occasional threads about Sabayon leeching (which it most certainly does).

There's just no interest with Gentoo, or not enough, on making it easy to install. Which is ironic, given that the whole *point* of portage is to make it easy to maintain!

Re:Trouble

[misleb]

I'm not saying that it is a defect of Gentoo that it isn't easy to install. The maintainers and users are free to make it whatever they want. I'm just saying that if most users are just going to copy/paste from a howto anyway, you might as well script it or something. A lot of people are spending a lot of time writing up explicit directions for things that could just be automated and/or abstracted. Which is ironic because the usually Linux is criticized for focusing on code and generally lacking in documentation.

Re:Trouble

[misleb]

This is what I did with my first Gentoo install. At first I had no idea what was going on. People that only follow the instructions never find out what's happening.

I've two very distinct modes of operation when it comes to directions. If there are no directions (or limited directions), I take it upon myself to find my own way and I learn a lot in the process. But when I am put in a position where I have explicit, detailed directions, my brain just kinda shuts down and I blindly follow the directions. Like if I am driving a car and I have someone telling me exactly where and when to turn, I stop thinking about where I am and just make the turns when requested... and I never remember how to get there on my own. But left to figure out it out myself, I retain much more. For this reason I'll never get a GPS for my car. If I did, I'd become so dependent on it, it would be pathetic.

Gentoo (the installation part, anyway) put me in a position where I really couldn't install it at all without following some guide. And the guides were there. It was just too easy to blindly follow them step by step. So i fell into that "mode."

Re:Trouble

[Anonymous Coward]

"Robins claims there was also a bug in a recent portage version that slowed things down quite a bit."

Robbins is more the expert than I am, but I think I'd have to disagree with him. I've upgraded portage several times since that post, and nothing has changed.

"Do you mean slow time to see updated ebuilds or that it takes a long time to compile?"

Where KDE is concerned, updated ebuilds was one of the problems I switched to Sabayon. Stable KDE ebuilds for Gentoo were still at 3.5.5, and Sabayon had stable ebuilds for 3.5.7.

I'm a gentoo fan, and I doubt I'll ever switch to something else like a debian or rpm based distro, but I believe that Gentoo is a mess and is being outclassed in a serious way by its own offspring, namely Sabayon, in much the same way that Ubuntu took precedence over Debian.

Re:Trouble

[turbidostato]

"Debian isn't without some weird issues either"

Of course not. That doesn't mean that anything you take our your a** is an issue.

"VNCServer doesn't depend on any font packages"

Of course not, since it doesn't really need any local font package installed. What else would you expect?

"despite not being able to start without them, with the official explanation being that the user could be running a font server somewhere in their network."

Isn't it a valid explanation? Wouldn't vncserver start using a network font server? And then, you see, both vncserver and vnc4server do *recomend* the installation of xfonts-base, just look at it.

Any modern OS is a complex thing and if you want to administer one of then, you'd better expend some time learning its ins and outs. That being said the "ins and outs" of Debian regarding systems administration are probably the very best out there.

Re:Trouble

[Anonymous Coward]

I'm curious to know which packages?

Re:Trouble

[Xabraxas]

It isn't Gentoo's "fault' that I didn't learn anything from it. I'm just saying that Gentoo elitists should give up the illusion that the install process is some profound leaning experience/initiation that every Linux user should have to go through.

I use Gentoo and I think that the install process is full of information. A stage1 install can teach you a lot about Linux like how to partition a disk in Linux, how to install a filesystem, how to install grub, setting up networking in linux, compiling a kernel, etc. You don't need to know anything about these things in most other distros to get the system up and running but it is required in Gentoo. There is good explanations in the hand book on how to do these things and what values you need to look for and use in your particular setup.

gentoo

[johnm1019]

I think gentoo has some incredible flexibility and it'd be a shame to see the project go by the wayside.

Re:gentoo

[Anonymous Coward]

So much for being "open" :)

Re:Gentoo

[Anonymous Coward]

That's hardly spam. They targeted a tiny set of linux users to see whether they want to try something else.

Because this worked so well last time

[Anonymous Coward]

http://lwn.net/Articles/224615/

So far it's looking good for him

[alveraan]

There's a sticky post in the gentoo forums dealing with this. So far Daniel got a pretty positive response and frankly... as a user that has seen gentoo slowly falling apart over the past few years, I'm glad he's motivated to bring it back on track: http://forums.gentoo.org/viewtopic-t-644321.html

Should we care?

[mustafap]

So a distribution might die. Good. There are too many of them anyway.

Re:Should we care?

[Nyago]

Except that Gentoo happens to be one of the best. Maybe if one of the dozens of Red Hat clones using the same crappy RPM system died, nobody would miss it, but... Gentoo is too important. Even the non-Gentoo users I know rely on the Gentoo forums and wiki and documentation for help.

Re:Should we care?

[Mantaar]

No. I think diversity in Linux is a Good Thing. There are hundreds of distros out there and that's really good to see, because they're all competing with each other, sharing their work with each other, forking one another and then merging back... If a distro dies, ten new ones spawn. That's very good, it contributes to a diversity which makes the Linux community an interesting place to live in.

And that 'but it confuses the newbies' argument just doesn't cut it anymore. For the complete boons, there's Ubuntu and probably SuSE. For everyone else, there's choice. I like choice. Right now I chose Debian, but that has changed in the past and will probably change in the future.

... as long as it's not RPM-based...

Re:Should we care?

[mustafap]

I disagree. In fact your point:

> There are hundreds of distros out there

is the bit that I would be concerned with. While I can't draw on any formal evidence, it seems obvious that there comes a point where diluting the development effort across an ever increasing number of distributions becomes counter productive.

I'll shut up now because this is hardly a new idea, but thanks for the debate.

Re:Should we care?

[jd]

The problem lies not with the number of distributions but with what the different distributions offer. Needs, and therefore "ideal" solutions, tend to be specialized. General-purpose distributions have to be generalized. This means that general-purpose distributions will meet most of most needs, but can never really be ideal for any of them.

Gentoo's approach of configuring and compiling at point of install should - in theory - solve this problem. You can adjust what gets compiled with what options and can therefore tailor the solution exactly to what you need. This is great for some of the more complicated packages, where there are many optional components, some of which may be mutually exclusive. This is also great when you have packages that - if you compile in everything - the package become unwieldy and sluggish.

In practice, the maze of options and the staggering number of potential compiler flags for tuning things -- it's simply too complicated for the majority of users and even for a very large number of software engineers. A better solution, in my opinion, is to have users describe a basic distribution and the platform on which it is to run, and then have a central cluster use herustics to grind out a way to achieve it.

Personally, I'd do this by compiling a mini distro locally that used a very standard package manager and didn't invalidate assumptions by mainstream distributions also using that package manager. Then the user could use existing repositories to add the stuff that's not critical to them but they still want. Alternatively, the cluster could spit out all of the necessary scripts, databases and configuration files for a Gentoo-style distro to build that perfect foundation.

However, ultimately, I do believe this to be the area virtually all distros get it wrong. The foundation components are the most critical, but they are also the least reusable. Correct that and you correct 99% of the (few) problems people have with Linux.

Re:Should we care?

[muuh-gnu]

> it seems obvious that there comes a point where diluting the development effort

This nonsense argument about dilluting effort gets repeated over and over. How exactly is it "obvious" that if I do something the way I think is the best, and work independently of the few "majors", I am dilluting the work of somebody else? Its called competition, and, as I least heard about it, it promotes diversity and is rather healthy for a ecosystem of any kind. There is no reason why there shouldnt be extreme diversity in the software landscape. You and your likes seem to suffer from some kind of software xenophobia. How exactly does some obscure source based distro you never heard of, make _you_ counterproductive? Using your reasoning Linux and the BSDs shouldnt even exist because theyre dilluting somebody else's windows based "development efforts".

Re:Should we care?

[Anonymous Coward]

>Using your reasoning Linux and the BSDs shouldnt even exist because theyre dilluting somebody else's windows based "development efforts".

Classic Straw Man Argument there. Well done!

His point is that if you have 100 developers each working on a distro, that's worse than having 100 developers working on a single distro.

Anybody remember MSX?

Exactly.

Re:Should we care?

[rjames13]

I think he is trying to say that if everyone for instance worked on Gentoo then work on Gentoo would proceed much faster. This thinking is flawed for another reason than what you pointed out. Distro's have a purpose, it may not seem like it to someone who sees 300 distros and thinks OMG that is way too many. But each Distro was designed to solve a problem, different from another Distro. If everyone worked on say Gentoo then only the Gentoo problems would be solved not the others. For example there are some Distros that are for Desktop use, some for Servers, some for partitioning, some for system rescue, some for High Performance Computing. Having one Operating System to solve all problems is a flawed idea, much better to break up Operating Systems into the groups of problems they solve, which is exactly what has been done.

Re:Should we care?

[hairyfeet]

Besides that, the great thing about the current Linux landscape is that most of the distro "trees" out there come from one or two "roots" which means that most of the software developed from those roots can be spread to the trees without duplicating effort. I run several Debian packages on my Xandros 4 Pro, as I'm sure that many Ubuntu and other Debian based distro users do. But I wouldn't have Linux if I had to run Ubuntu(can't stand it) or Red Hat(REALLY can't stand it) just as I'm sure that Ubuntu and Red Hat users would hate it if they had to all switch to Xandros.

As I've said before, all we really need is an easy to use website with side-by-side comparisons of features along with a way to search and see which ones support your particular hardware "out of the box", so folks can find the right one to fit their needs without having to try dozens like I did.

For me, Xandros just works. For you, it might be Debian, Ubuntu, or one of the myriad of other flavors. We simply need a central place you can check out the different choices and find what is right for you without downloading every distro on the planet. But I think Vista is a good example of why choice is a good thing. With Linux you can choose to be cutting edge and have the bling, or like Xandros be rock solid and just work out of the box without the latest effects, or even build it yourself like Gentoo. Choice is one of the great things about OSS IMHO.

Re:Should we care?

[wanderingknight]

A sibling post already nailed it, but summarized, the point is as follows: It's fucking open. The regular secrecy rules of capitalist competition don't apply here; 90% of the time, changes are forced to be made open (obviously, depending on what license we're talking about). Any change in any distro is open for any other distro to grab and to use. And since every distro, in the bottom, is the same fucking system, only that delivered in a slightly different way, any progress of any distro is good for the whole community.

Re:Should we care?

[Anonymous Coward]

Doesn't work by choosing hardware specs, but http://www.zegeniestudios.net/ldc/ claims to help you find a distro based around what features you need.

Re:Should we care?

[Jesus_666]

Actually, Gentoo is one of the two distros I see most commonly offered to newbies, the other one being Ubuntu, of course. Those newbies who want an "it just works" distro get Ubuntu while those who have specific needs such as a shell-only test box on old hardware are told to get Gentoo. Even if Portage is getting slow, it's still so much more useful than everything else out there that most advanced users I know tend to gravitate towards it because they are averse to the ridiculous dependency graphs binary -ased package managers tend to generate.

Also, Gentoo is the best-documented distro out there with Gentoo HOWTOs often containing very useful information even for non-Gentoo users. It's pretty much irrelevant what you intend to do on your Linux box, a google for [subject matter] gentoo will usually give yu a detailed description of what you need to do.

Gentoo is much more than the ricer distro many people see in it.

Re:Should we care?

[Anonymous Coward]

"This is also great when you have packages that - if you compile in everything - the package become unwieldy and sluggish."

Does this actually happen?
I used to use gentoo for quite the opposite reason to most people. Not to get a minimal system, but so I could get many of the obscure options that packages for other distros leave out. (Mainly for graphics/media/audio stuff)
Then I'd enable a load of other flags just in case I forgot something.

I did not find that truly awesome amounts of 'use flags' made any difference to how fast the packages ran, or how speedy the computer felt.

Re:Should we care?

[hairyfeet]

Thanks but, as you ponted out it doesn't include hardware and its idea of "help" is a very lame question and answer. I purposely picked answers that should have skewed it towards Xandros (which is one of their supported distros) and instead was told my best choices were Ubuntu(hate) OpenSUSE(Yuck) and Kubuntu(big difference between that and Ubuntu,if it doesn't work in Ubuntu,Kubuntu isn't going to either).

That is EXACTLY why we need a site that includes hardware into the equation, and not just asking you lame questions. If I literally wouldn't have had a Xandros 3 Pro box set dropped into my lap by a SUSE fan I would have given up on Linux after more than 3 dozen different distros failed and stuck with XP, and by doing so I wouldn't have learned enough to give those who couldn't afford a pc an older refurb with a low resource distro. As most of the places that donate machines to me take the CALs, I simply would have stripped them for parts and tossed them out.

The reason I learned enough to begin to DIY is because of trying new things on Xandros and learning as I went. Package management, mixing different repositories, compiling from source, etc. Most folks just aren't going to have the patience to try dozens of different distros, nor or they going to be as lucky as I was and have the perfect distro for them dropped in their lap. This is why a central database with features, screenshots, and most importantly tested hardware supported side-by-side would IMO increase Linux adoption by an order of magnitude.

Look at how many machines out there are being shipped with only Vista drivers but don't have the power to run more than basic. With an easy to use website that would suggest based on hardware and features required a half dozen or so distros based or relevancy to start with, along with listing the paid alongside the free distros would help a LOT of folks switch.

Sorry for the length, but this is a topic I fell VERY strongly about. I have had many folks that I could have switched, but ran into a single piece of hardware that wouldn't play nice and had to give up after many fruitless hours searching forums. If I would have had a central plac I could have checked to see if their Compaq, Emachine, Dell, etc were supported and by what distro I could have saved a lot of time and those folks wouldn't be dealing with Windows virii right now. Now if we could only add to that an Ndiswrapper for those damn Lexmark printers my wishlist for Linux would be complete. ;-)

Re:Should we care?

[Richard_J_N]

Agreed. The gentoo docs are the best for a lot of things. Also, gentoo's boot system is by far the prettiest and clearest of any I have seen. Unfortunately though, when I tried the installer, it crashed out, and I didn't have the time to get everything fixed - so back to Ubuntu and Mandriva.

FWIW, I'd love to see Gentoo and Mandriva merge. They are both great projects, and both slightly too small. Also, they have different aims, which would dovetail quite nicely...

Re:Should we care?

[Anonymous Coward]

Correct that and you correct 99% of the (few) problems people have with Linux.

So, you're saying there are at least 100 problems? Or could one have a fraction of a problem?

Re:Should we care?

[jd]

You can have fractional problems on a Pentium I.

mod fuP

[Anonymous Coward]

FrreBSD 4roject,

Gentoo as a learning aid

[dmneoblade]

Gentoo was my way of learning a lot about linux sysadmining in a short time. In a couple weeks, I learned how to compile packages, manage partition issues, compile kernels, deal with numerous config files, and many other skills. I later switched to Ubuntu, but I still appreciate my time spent with gentoo as a great learning aid. Just enough help to make it not as hard as LFS, but hard enough to be challenging.

Re:Gentoo as a learning aid

[matria]

Then you should really appreciate the LFS program...http://www.linuxfromscratch.org/

Re:Gentoo as a learning aid

[jawtheshark]

... Which he mentions in his own post ...

Re:Gentoo as a learning aid

[Teppic_52]

Quote from the Gentoo forums

I would have used LFS, but didn't fancy using a notebook as a package manager

Re:Gentoo as a learning aid

[rjames13]

What's wrong with make uninstall? It works...sometimes :)

Re:Gentoo as a learning aid

[Sancho]

Package managers USUALLY let you remove dependencies that are no longer required, too.

Re:Gentoo as a learning aid

[debatem1]

Absolutely. The highly technical nature of Gentoo appealed to me right from the start, and so it was the first Linux system I played with. Just getting it running was several days of ignorance and fdisk, but by the end I knew quite a bit about the command line and Linux internals. Fast forwarding to today, I would say that it is a rare day that goes by in which I do not use something I learned in that first gentoo experience.

Re:Gentoo as a learning aid

[buchanmilne]

Well, I believe many of the users who claim "Distro xxx taught me nothing, then I tried Gentoo" just didn't bother learning anything on distro xxx. It's not as if there is one Linux distro available where you are prevented from learning something. And while Gentoo users claim they learnt so much ... most only know how to set USE flags, use emerge, and install a new kernel from source. Simply going through something like "Securing and Optmizing Linux" will teach you more on any distribution (not that I agree entirely with the contents myself).

Since I haven't used Gentoo myself (just helped many Gentoo users get it installed, or fixing some problem preventing them from using Gentoo, or setting up some specific feature), I may be biased, but I personally feel Gentoo is a waste of time.

Re:Gentoo as a learning aid

[debatem1]

Its the installing, not the using. It can be as easy to use as any other distro once installed, but it is- depending on your perspective- either a pain in the ass or a great learning tool to install. Pretty much like LFS, which I also think is a great learning tool. Bottom line is, I can really only learn this sort of thing by doing, and it was very helpful to me. Your mileage may vary.

Gentoo needs drobbins

[Reverse Gear]

The reason for this offer from Daniel is imho not as important as it is that he is offering to step up back as the leader of this project and take his job down to part time so that he again can put some energy into the role as leader of Gentoo Linux.
Gentoo badly needs some leadership right now, Daniel has done it well before and had Gentoo thriving while he still was at the helm, so let's get him back there and get this ship back on course.
I really hope that the council will accept this offer for the best of Gentoo and not let their personal agendas stand in the way of the good of Gentoo ... though I fear that might happen once again.

His post said

[joeflies]

that he wants an answer in 7 days. There's no way that your $emerge leadership package will compile and install by then.

Leadership: Gentoo-way

[Dark Coder]

#
# emerge -C gentoo-leaderships
# emerge -uDv gentoo-leaderships
# echo "Deep Leadership Upgrade: Done."

Here's to hoping that its "package" dependencies don't break.

Why is the foundation required?

[grahammm]

I have been a user of Gentoo for time and have never seen an explanation of why the foundation is needed or even what it does. Looking at the home page under 'About Gentoo', 'Philosophy' and 'Social Contract', I do not see a foundation mentioned at all. To quote from 'About gentoo'

To advise on and help with Gentoo's global development, a 7-member council is elected on a yearly basis which decides on global issues, policies and advancements in the Gentoo project. . To my mind the council seems to be sufficient, so why the need for and fuss about a(n almost unmentioned) foundation?

Re:Why is the foundation required?

[Rakishi]

The foundation is from what I can tell the LEGAL entity behind gentoo so the domain belongs to the foundation, for example.

Re:Why is the foundation required?

[Teppic_52]

They own the IP too, all the Gentoo specific configs have their copyright notice in them.
I believe they control the bank account too.

Re:Why is the foundation required?

[grahammm]

They own the IP too, all the Gentoo specific configs have their copyright notice in them. But all of the IP is licensed under GPL-2, so anyone can use, modify and/or distribute it subject to the GPL-2 rules. I know that where a 'natural' person owns copyrights that if they die then the copyright passes to their heirs. What happens to copyright which is owned by a corporate entity when that entity is dissolved?

Re:Why is the foundation required?

[Teppic_52]

What happens to copyright which is owned by a corporate entity when that entity is dissolved? I believe that to be one of the current problems.

Re:Why is the foundation required?

[greenrd]

If someone buys the copyright (remember an entity going bankrupt will sell assets to pay creditors), they get it. If no-one buys it... who's going to sue? It's effectively public domain, right?

Re:Why is the foundation required?

[gentooligan]

Not all IP. The Gentoo art work (logo's etc) and the Gentoo trade mark are not GPL.

Re:Why is the foundation required?

[AndrewM1]

Legally, the foundation is a registered nonprofit, which allows them (through the arcane tax code of the United States) some benefit in dealing with donations. Tax receipts, I believe?

Re:Why is the foundation required?

[nuzak]

Debian seems to do fine with just one elected board. I dare say a lot better than Gentoo, in fact. Possibly because if someone doesn't do their job, they're out.

And Robbins? Isn't he the same guy who returned to Gentoo like MacArthur to the Phillipines, only to leave one day later? I agree that the Gentoo board needs replacement too, but he ain't the guy for it either.

Same here

[theurge14]

I left Gentoo for FreeBSD due to these reasons and also due to waiting for certain packages for too long, then receiving buggy packages and finally, having the base config change several times in 6 months, mainly for apache2, php, etc. After spending a week with FreeBSD I don't think I'll be back to Gentoo for any reason.

Re:Same here

[Zarhan]

That's funny, I left FreeBSD for Gentoo for exactly the same reasons (this was around time of FreeBSD 5.0, so I probably had the worst possible FreeBSD experience). And Linux kernel had better hardware support, especially for laptops.

With FreeBSD, packages tended to break with almost every upgrade. With Gentoo, they still break after every upgrade, but at least there is revdep-rebuild to fix things. Portupgrade -L didn't really work...

Re:Same here

[rainer_d]

FreeBSD is not without problems, but I still don't see a replacement on the horizon any time soon.
And FreeBSD does have a leadership.
If the ports-system would work 105% on (Open)Solaris, I'd change ship for most bigger servers. As it stands, a lot of stuff that doesn't involve a complicated software-stack is going to make this switch in the mid term here. Provided, I can streamline the installation enough and we get the patching process under control.
But anything with weird PHP dependencies (and that includes, strictly speaking, phpMyAdmin) is a real big time-killer on anything else but FreeBSD.

Re:Same here

[Anonymous Coward]

Gentoo was an amusing fad. I remember how many Slashfolk raved about it like it was the second coming of Christ. Now y'all are kicking it in the teeth.

Re:Same here

[misleb]

With FreeBSD, packages tended to break with almost every upgrade. With Gentoo, they still break after every upgrade, but at least there is revdep-rebuild to fix things. Portupgrade -L didn't really work...

Heh, reminds me of what happened on my FreeBSD mail gateway running amavisd and spamassassin. You know how those two packages alone require like 1,645 different perl modules? Well, I made the mistake of updating perl from ports. It happened to be a major version upgrade and paths changed... and every single perl module on the system was hosed. Long story short, I ended up just reinstalling FreeBSD from stratch rather than wade through the tangle of perl module dependency hell. Got a more recent version FreeBSD in the end, but man did I discover the drawbacks to not having a stable and well tested package set (as in Debian, for example). Ports (and portage for that matter) are nice if you always want to keep recent packages, but managing it all really can be a lot of work.

Re:Same here

[setagllib]

NetBSD's pkgsrc project (which works on FreeBSD and Linux) solves this by having stable branches much like Debian, but once per quarter. So you can keep a stable package tree with security fixes, but still compile from source with your custom configuration. There are other problems like very hackish upgrading, but at least that part was done right.

Gentoo

[tiny69]

Troll mode on:

Gentoo has always been a group of amateurs. When Gentoo first started, they'd SPAM Slackware IRC channels. "You know how to compile software. Do you want to become a developer for Gentoo?"

Just look at the history of problems they've had with emerge or whatever they call it. /me ducks

no. Gentoo is different enough

[alizard]

at a technical level that it's worth saving. Someday, distributing apps as binaries may hit a wall in terms of efficiency and matching individual hardware and if that day comes, we'll be glad there's another way to do things out there which has people who understand it.

The state of Packagage Managers

[Zombie Ryushu]

There is but one Linux. But there are multiple forms of packaging. Mandriva and Fedora/RedHat/CentOS can be lumped into the category of having dependancy resolution problems too rigid. Mandriva specifically suffers a loss of redundacy when a source of RPMs fail. Yum keeps mirror catalogs. urpmi from Mandriva has the added ability of using SSH and Kerberos to "Mass Deploy" applications, and can centralize with LDAP. No other package Manager can do that.

But both urpmi and yum fail at handling source code package. You have to download them and compile with rpm --rebuild.

When it comes to Debs, I have no idea how to build Debs. Ubuntu and Debian lack the SSH/Kerberos mass deply ability and are even HARDER to recompile than RPMs.

Portage Handles source Code gracefully. Thats its strength.

I sorta wish that some of these projects would merge. I wish that urpmi handled mirror failures and did a better job than it does. I wish that urpmi could handle source code. Likewise, I wish that yum could use SSH/Kerberos

Now, Slackware tends to be problematic, no package dependancy can result in chaos.

I also take issues with things like Autopackage RPM and DEB are here to stay. get over it.

Re:The state of Packagage Managers

[Yfrwlf]

I wish everything was modular like it should be, so "Linux" was a single "platform", and that it didn't matter what distro you were running because it was merely an certain selection of packages and nothing more, and if you liked the good things about one distro but not others, you could install those good things on your own distro.
 
Linux desperately needs modularity and a universal package management system, so it won't fragment and die like most of Unix.

Re:The state of Packagage Managers

[BlackCreek]

When it comes to Debs, I have no idea how to build Debs. FYI: I think the ``Correct Way'' is: 1. Either download directly or use apt-src 2. then: dpkg-source -x $package_name.dsc cd $package_name dpkg-buildpackage -rfakeroot Or you could *try* to use apt-build (but I don't know if they ever got that to work as advertised). In a hurry you could also use "checkinstall" http://asic-linux.com.mx/~izto/checkinstall/ (which is a sort of hack for packages without the files to use the first two suggestions). Cheers,

Re:The state of Packagage Managers

[BlackCreek]

> When it comes to Debs, I have no idea how to build Debs.

FYI:
I think the ``Correct Way'' is:

1. Either download directly or use apt-src
2. then: dpkg-source -x $package_name.dsc
cd $package_name
dpkg-buildpackage -rfakeroot

Or you could *try* to use apt-build (but I don't know if they ever got that to work as advertised).

In a hurry you could also use "checkinstall" http://asic-linux.com.mx/~izto/checkinstall/ (which is a sort of hack for packages without the files to use the first two suggestions).

Cheers,

Re:The state of Packagage Managers

[rjames13]

Now, Slackware tends to be problematic, no package dependancy can result in chaos.

Yes and us Slackware users divert that chaos through /dev/random increasing our cryptographic key generation abilities.

Seriously I have used Slackware since before ver 3 and have never seen chaos from dependancy issues. You make it sound like it crashes computers at random. But as someone who actually knows how it works I can tell you this, all a unresolved dependancy issue does is stop a specific program from running until that dependancy is met. If foo needs bar then the system does not crash foo just complains and exits. This is Linux not Window95.

Re:The state of Packagage Managers

[sydneyfong]

[Disclaimer: I am only a casual tinker of debs, so my information may not be all that accurate.]

When it comes to Debs, I have no idea how to build Debs. Ubuntu and Debian lack the SSH/Kerberos mass deply ability and are even HARDER to recompile than RPMs. The Debian way to "mass deploy" packages is probably using the apt-get system. You simply add your own package repository to your "sources list", and use apt-get to install/upgrade the packages. They do provide gpg signing, if you want security.

Other posters have mentioned some possible ways of building from source. For me, at least if the source package is from Debian official, it's a matter of `apt-get source PACKAGE`, then `cd DIR; fakeroot debian/build binary`. If you have the prerequisite development packages install the package should build cleanly and you'll have the newly compiled binary packages ready for installation.

[By the way..... claiming that building debs from source is "harder" because you "don't know how to build Debs" is not terribly convincing...]

Re:The state of Packagage Managers

[MasterOfMagic]

To rebuild a Debian package from a source (as opposed to a binary) repository:
apt-get source <pacakge-name>
cd <pacakge-dir indicated in the output from apt-get source>
dpkg-buildpackage -rfakeroot

Done. The process will spit out a deb you can install with dpkg. There are a few open source projects (mplayer, vlc, quakeforge) that add the debian directory in their mainline source. After extracting a source tarball from them, you can run this command in the root of their source tree and get .debs.

This has the nice benefit of also checking to see if you have the proper build dependencies to build the package as the package creator saw fit. You can hack the debian/rules and debian/control file to customize how the package is built (debian/rules is a shell script and debian/control follows a standard format). You can override this dependency check with a command line switch to dpkg-buildpacakge.

Re:The state of Packagage Managers

[jumperboy]

Now, Slackware tends to be problematic, no package dependancy can result in chaos.

This is not true in my experience. Slackware has legendary stability, partly due to not having a package manager that tracks dependencies. If you're missing something, just install it (however you want) and move along. Try repairing a dependency-based package manager when it becomes corrupt -- now, that's real chaos!

Re:The state of Packagage Managers

[domatic]

I've built several things from Sid on Ubuntu Gutsy by following these instructions from Debian's FAQ:

7.13 How do I install a source package?

Debian source packages can't actually be "installed", they are just unpacked in whatever directory you want to build the binary packages they produce.

Source packages are distributed on most of the same mirrors where you can obtain the binary packages. If you set up your APT's sources.list(5) to include the appropriate "deb-src" lines, you'll be able to easily download any source packages by running

          apt-get source foo

To help you in actually building the source package, Debian source package provide the so-called build-dependencies mechanism. This means that the source package maintainer keeps a list of other packages that are required to build their package. To see how this is useful, run

          apt-get build-dep foo

before building the source.

If you want just to compile the package, you may cd into foo-version directory and issue the command

          dpkg-buildpackage -rfakeroot -b

to build the package (note that this also requires the fakeroot package), and then

          dpkg -i ../foo_version-revision_arch.deb

to install the newly-built package(s).

Re:The state of Packagage Managers

[domatic]

I've had multiple Debian and Debian derivative installs last through years of my installing, removing, trying things out, and just generally being mean to the system. I've had to fix minor package manager faults but that wasn't anything like the trouble I've had manually managing dependencies and cruft on non-Debian based systems. For the most part, Debian's package maintenance systems Just Work.

Re:The state of Packagage Managers

[celle]

Slackware has dependency management, its call the system administrator. It means you better know what your doing or it may bite you. Please refer to Patrick's message in the early releases about slackware being for programmers(precursor to carmeggedons non-apology for gore). Besides it would help lots if application programmers would document dependencies for their application properly instead of installers(the human kind) having to guess at dependent apps and their configurations. To your credit, it has gotten better the last couple of years. Many more programmers are documenting their code and doing a better job, but there's still a lot of variation in standards of what to document. Autoconfig makes a big difference and helped alot but dependency configuration still can be hell if the dependent programs used a configuration that is not the default. Please list when you change it so we have an idea.

Ydrassil early 1990's. Slackware user since 1994?5, its been that long? Debian off and on since 1997, freebsd early 1990's and since 2004. Various others I just can't remember them all any more.

Re:The state of Packagage Managers

[buchanmilne]

The Debian way to "mass deploy" packages is probably using the apt-get system. You simply add your own package repository to your "sources list", and use apt-get to install/upgrade the packages. They do provide gpg signing, if you want security.

And exactly this is usually accomplished under (say) Mandriva using urpmi/rpmdrake, by adding your repository with urpmi.addmedia. However, this is not the mass-deploy aspect referred to, which on Mandriva is provided by urpmi-parallel (to install/upgrade/remove the same set of packages on a group of machines). Similar functionality exists for Red Hat from RHN, but I'm not aware of any other distribution having a system like this.

Re:The state of Packagage Managers

[buchanmilne]

But both urpmi and yum fail at handling source code package. You have to download them and compile with rpm --rebuild.

If you have the source repos enabled, you can use urpmi --install-src to download and install the srpm for you, and urpmi -s to install the build requires.

The thing is, if you're going to download an srpm, chances are you want to make some change to it. If not, why don't you install the available binary, or request a backport to be shipped?

For package maintainers, there's no benefit to being able to install from source ... as in the end the package maintainer has to support the binaries, not just the source, and the existing package maintenance tools (repsys, mdvsys etc.) which allow easy use of the svn repo, submitting packages to the build cluster etc. allow for convenient package maintenance without ever seeing a SRPM.

While a "install the latest from cooker on my ancient install by recursively rebuilding the build requires" feature has been discussed in the past, it was instead decided to put more emphasis on backports, which (assuming popular packages are made available for at least the current stable release, in some cases a few releases back) satisfies what probably >90% of users wanted anyway.

Re:The state of Packagage Managers

[sydneyfong]

how about:

for host in LIST: do
    ssh $host apt-get install PACKAGE
done

Or maybe having a cron job check your custom repository for updates and automatically install stuff?

Re:The state of Packagage Managers

[Ash-Fox]

When it comes to Debs, I have no idea how to build Debs. Ubuntu and Debian lack the SSH/Kerberos mass deply ability and are even HARDER to recompile than RPMs.

sudo apt-build install applicationname

It will download the source, compile it, package it and install it.

Unless you meant actual packaging, then you need to follow the Debian New Maintainer's guide.

Re:The state of Packagage Managers

[Ash-Fox]

Linux desperately needs modularity and a universal package management system, so it won't fragment and die like most of Unix.

LSB dictates RPM support for it's universal package support, as far as I know, most popular distributions are LSB compliant, even Debian an Ubuntu are.

Gentoo on the other hand being LSB compliant depends on how you configure and setup everything.

Re:The state of Packagage Managers

[Yfrwlf]

I'm talking about proper support, not theoretical support. The fact is installation often fails a lot of times when trying to install packages outside the repos. There needs to be a universal format that just works, like Windows and Mac have. I wouldn't mind if there were multiple formats, that's fine if there are RPMs and DEBs and Autopackages competing for the spotlight, competition is vital, and you can have that competition if you have standard APIs with the package managers for package formats, but each of those formats need to be able to correctly install themselves on LSB-compliant systems before non-repo software installation as as easy as the aforementioned OSes.

Re:The state of Packagage Managers

[Ash-Fox]

I'm talking about proper support, not theoretical support. The fact is installation often fails a lot of times when trying to install packages outside the repos. There needs to be a universal format that just works, like Windows and Mac have.

There is. RPM on LSB support distributions (I use LSB applications, so this is not theoretical like you claim), Loki installer among others. LSB RPMs have never failed for me.

but each of those formats need to be able to correctly install themselves on LSB-compliant systems before non-repo software installation as as easy as the aforementioned OSes.

There is only one package format considered to be LSB compliant and that is RPM. Provided the applications follow the LSB, there is no problem.

If you're going to complain about package incompatibility from other distributions which are compiled against a specific distributions, this is exactly like complaining when games compiled against the latest Window XP SP2 platform SDK don't work on Windows 2003 or Windows XP (no service packs).

Re:The state of Packagage Managers

[Yfrwlf]

Not just me complaining about it.

http://www.linux-foundation.org/en/Packaging

There is a problem with the current state of packaging in Linux. It shouldn't be so difficult, not only for end users but for developers. Common APIs need to be created so that packaging formats can compete and developers and users can have easy access to a simple system that does the things they need. OK, so RPMs work, why is it then that they are often not available? Maybe it's a format that could use some improvements, but regardless of the ease of use or power of RPMs, depending on one type is silly. The fact is, RPM isn't as common as it could be, regardless of the reasons for that, and that's a pretty critical part of the "Linux platform" for developers wanting to port or create programs for all LSB-compliant distros and for users wanting to just get to their programs without any fuss.

What defines a system should be the APIs so that you have modularity instead of being locked into one kernel, program, library, or format type.

Re:The state of Packagage Managers

[Ash-Fox]

OK, so RPMs work, why is it then that they are often not available?

In the instances where it isn't available, I believe it's due to the fact that the developers try to support distributions beyond the ones which are just LSB compliant (beyond Gentoo.. Is there even a major distribution that isn't LSB out of the box?). StarOffice and Unreal Tournament certainly try this.

As for opensource software, most of the time it's been packaged for specific distributions and they assume any distribution that wants this software can either take their source packages and compile them or use their unique packaging format.

LSB compliant packages have a issue of taking up larger amounts of space, since many libraries outside of the standard base are supposed to be packaged with the application, so it works independently on them. Requiring more memory in RAM (as they aren't shared libraries) and harddrive space and this is likely one of the main reasons why opensource software is not usually in a generic package.

What defines a system should be the APIs so that you have modularity instead of being locked into one kernel, program, library, or format type.

You aren't locked into doing anything on Linux. You don't have to follow any of the standard practices either. It's a double-sided sword. Of course there are always issues to agree upon when it comes to standardization and I don't really see a problem with packaging currently. It isn't like OS X 10.5 or Windows Vista where UI features, standards are written in stone and you should not deviate from and in some cases, cannot deviate from.

Re:The state of Packagage Managers

[Yfrwlf]

So you're saying this term isn't real because you ultimately always have a choice http://en.wikipedia.org/wiki/Vendor_lock-in, I should stop bitching about laws in the US because I have the choice to leave this country, and you should stop worrying about curing cancer because anyone with it has the option of suicide. My point is that you missed the point. It's not white or black. Things can be made better than they are now. But you are aware that software packaging is a problem apparently, which is good, because that's the only point I'm trying to make here. Whether the answer is as extreme for everyone using Ubuntu and Debian to suddenly all switch to Red Hat, Fedora, or CentOS, or the answer is for those distros to start using RPMs, or (a much better solution) a common API for the package managers is used so that you can install both RPMs and DEB packages with no problems out of the box (I hope the Linux Foundation can help to come up with a solution), something DOES need to happen, because it IS a problem for lots of reasons that I haven't even mentioned yet and if you read that link I sent you you'd read about all the things that are needed but aren't provided currently. I'm trying to give you the hint though that being thick-headed and ignoring that there are current problems with it isn't the solution. I can think of lots of features I want to see added onto a packaging format, and I want to see the formats compete, so I can't wait for a new API allowing this to happen. Also I'd like to see one extension, .LNX or something, to be created as the container format for these formats, then having RPMs and DEBs be sub-formats, among other things. I have lots of other ideas too, there are lots of things that are needed here to make packaging in Linux a snap. :)

Also, didn't you heavily contradict yourself? You probably think that OpenGL is great, and you've already proven to me that you believe standardizing on RPM and the LSB is what we should all use, so you believe in pushing standards apparently, yet you seem to believe that any standards means somehow Linux is no longer free and that we're all in some prison where we're forced to comply. That's messed up, and backwards. Having standards gives us the freedom to be able to share things and communicate on the same frequency. You are using countless standards as you read this, from HTML to other web compliance standards to the software you run and the APIs programs use to communicate with each other, down to the hardware and the common standardized slot types and architecture standards, and even lower to the standards of electricity so that all the playing fields are level and things can compete in the same arena. That's what big corporations don't want. Do you think Pepsi and Coca Cola want you to be able to choose between them? No, they want to pay off the company owning the land and have them sign exclusivity agreements so they can eliminate the competition and thus drive up prices and make a huge profit. Do you think Microsoft wants other operating systems in stores next to theirs, or for computers to come without their OS forced down the throat of consumers, not even having the ability to have the software removed so they don't have to pay for it? I don't mean to be rude, but if you think having standards and a level playing field for competition is bad, you need to wake up. Yes, of course some "standards" could use an improvement, and you know the best way of doing so is by pulling back one level and allowing something to simply be compliant to a standard, so that the competition has room to step forward, and then competition can occur like it should be able to.

There are literally a million examples I could give you, so I'll stop there. Linux standards are good, allowing users to share programs easily that they have created. Right now, that is not done in Linux as much as it should be and there are reasons why and lots of

Gentoo & OpenEmbedded are Unique Linux Dists

[Anomynous Coward]

The form of 'metadata' stretched out on top of both these distributions are what make them unique.

portage for Gentoo & bitbake for OpenEmbedded

With each, a (usually) high degree of control is given in how to shape the distribution's function and attributes. And this is repeatable from "first principles" compilation of source code (from tool chain onwards).

What is needed is research and standardisation in the ontology of this metadata and it is for this I believe Gentoo can still play an essential role.

Anyone who has spent a lot of time tweaking parts of their application, OS space, kernel and boot methods would certainly appreciate the ability to reproduce that work from scratch if need be.

To publish and share this metadata distribution 'state' is to fine tune the virulence of GNU/Linux beyond the GPL and into the real (and virtual) system space.

shine, .vortex

happens to copyright?

[Anonymous Coward]

This is well defined under the incorporating state law. They pass the IP, liabilities, money and other property to another willing entity incorporated under the same clause.

501c3 -> 501c3. Happens all the time. Though the legal negotiations usually last until bank accounts approach zero.

Re:happens to copyright?

[LingNoi]

Any code changes are licensed under the GPL, your state is welcome to make changes to their GPL'd code and keep it secret, that is if they can find which lines were edited. Otherwise they might be violating someone elses GPL'd code which would mean they lose their right to use that persons GPL code that was included with theirs rendering any of their newly acquired code useless.

Since a lot of upstream projects request that you assign over your copyright to the project's admin, actual code owned by Gentoo could not even exist.

Only upstream matters

[LingNoi]

Except it isn't a good point at all because all the developers work together anyway regardless of which actual distribution they are working on.

1) Developers of the different distros contribute all their changes back to upstream. It doesn't matter how many developers are working on which distribution because the fixes all end up back in the same place.

2) Not only this but the different distributions are all going in different directions and have different goals. When something is successful they'll borrow from each other.

3) Ubuntu's goal is to bring Linux to the masses. Debian's goal is to provide a libre operating system. There is no way you could put both developers of those groups together on a single distribution because they both have different goals, but it doesn't matter because they both share from each other and fix problems.

4) A bug in Ubuntu could be a bug in Debian which could be a bug in every distribution. What is most important though is upstream which is where all the changes take place. You seem to have made the mistake of thinking that every distribution has to code all the upstream fixes themselves when what actually happens is that they pull down the latest updates on all these projects and spend a lot of their time making sure everything in the distribution works. If it doesn't file a bug upstream and get it fixed. Hey, now it works on other distributions too!

5) So as you can see everything is co-operated on, even proprietary programs. While reading the uvc webcam driver mailing list yesterday a developer filed a bug in the Skype beta bug tracker as he found a problem with Skype's webcam implementation and even offered to co-operate with them on how to fix it.

Re:Only upstream matters

[mustafap]

>Except it isn't a good point at all because all the developers work together anyway

In some cases, yes. But is that true of the majority?

    Yes, if we are talking about drivers.

    Are KDE and Gnome working together?

My experience of managing an open source project which forked several ways was that the dilution of effort *did* affect other peoples work. 'Camps' appear, with work being duplicated simply because people would not take the effort to work together. You only have to look at how I was marked a 'troll' earlier on to see how difficult it is for some technical folk to consider a reasonably stated argument. Why debate something when you can fork, and ignore the people you disagree with.

Of course we need diversity. But do we need so much?

Re:Only upstream matters

[TheLinuxSRC]

Wow... some mod has your number...

Back on topic; I take exception to the example you give. Not only do KDE and Gnome work together, but KDE and Gnome both have their respective places. I know Gnome is more popular with most of the big-name distros but that does not change the fact that I cannot stand it :) I have tried to like Gnome many many times however its complete lack of configuration options quickly drives me up a wall.

I cannot stand gterm (or whatever the default terminal in Gnome is) compared to konsole. As an example, in konsole when you have multiple tabs open switching between tabs is as easy as shift-(right or left arrow); in Gnome's terminal it is some esoteric 3-button key combo. I also like having Konqueror set up with the tree view on the left and the konsole terminal at the bottom in the file browsing mode; though this is no longer available in Dolphin which I see is poised to replace Konqueror for file browsing :(

Having said this, I can see where some of the options are overwhelming to the new KDE user and I find myself setting up Gnome for new Linux converts rather than setting up KDE. This is because I am generally setting them up with Ubuntu and 90% of the posts on the Ubuntu forums assume you are running Gnome. The funny thing is, most people who see my KDE desktop after they have used Gnome for a while will usually ask about setting theirs up in a similar manner at some point. I cross that bridge when I get there though.

Re:Only upstream matters

[mustafap]

> Wow... some mod has your number...

That's slashdot for you. :o)

It's interesting what you say about KDE & Gnome, thanks.

Re:Only upstream matters

[ultranova]

As an example, in konsole when you have multiple tabs open switching between tabs is as easy as shift-(right or left arrow); in Gnome's terminal it is some esoteric 3-button key combo.

CTRL-(PageUpor PageDown), actually. Opening a new tab is the three-button combo (SHIFT-CTRL-T).

But, to stay on topic: I've personally found Linux From Scratch extremely helpful in updating and maintaining my RH9-based chimeiresque computer, but Gnome have pretty much given up in getting Gnome work properly; there is just too many interdependencies between it and various other packages. I wonder if I should try KDE at some point, or go with Enlightenment... then again, Ratpoison would give my poor mousearm a break :).

Suffered more than that ...

[Anonymous Coward]

I was a long time Gentoo user who got fed up and move on. As a software developer myself I found it too laborious to figure out why the latest emerge brought my system down. QA was non-existent and especially so for x86_64 platforms.

Gentoo overall is a decent idea for people who like configurability, but it's too fragile. For instance, I recall one update to the latest glibc which had some bugs in the portage script. Next thing you know nothing, and I mean nothing, works, since they all link against it. There is no "roll back" ability in gentoo, which honestly would be a good idea (and given they have installed files lists it's entirely possible).

I switched to Ubuntu and haven't looked back. While it's less configurable than Gentoo it's more than good enough, and it "just works."

In reality, Gentoo needs to improve the QA and safety (e.g. roll backs) facets more than their leadership.

will fail silenly

[tiq]

You wan this I think:

emerge -C gentoo-leaderships &&
  emerge -uDv gentoo-leaderships &&
  echo "Deep Leadership Upgrade: Done."

Re:will fail silenly

[ettlz]

Who cares? The GP's script was just three lines of comments, anyway.

Please make a donation.

[Anonymous Coward]

www.freerice.com

No thanks, Ubuntu is king

[Anonymous Coward]

Say what you will about Ubuntu, at least Mark Shuttleworth didn't kneel before Microsoft.

Gentoo

[Anonymous Coward]

I've was using Gentoo since pre version 1.0, I've submitted ebuilds that got accepted in portage and was a contributer at heart. I noticed a big change when Daniel Robbins stepped down, a big enough change to get me to drop the use of Gentoo.

I would love to welcome Daniel Robbins back and and I wish there was a way to allow community vote.

I gave up on Gentoo

[Anonymous Coward]

Leadership was a mess and it trickled down to the regular users. Fortunately, I found a system that lets me be even more elitist and I haven't been happier. OS X is great, and my new octo core Mac Pro is AWESOME! I can download mp3s really quickly now.

Re:Gentoo & OpenEmbedded are Unique Linux Dist

[Anonymous Coward]

I have a nasty stack of APR / Python / Postgres related packages
and I need something better than my own cobbled together build
scripts. I'm not interested in anything that forces me to rewrite
those packages makefiles or config scripts - I just want something
that allows me to easily maintain the depenendency tree and occasionally
apply a patch or two.

From what I've seen bitbake is really nice for this but there's
this enormous metadata thing you need with it and the docs for
bitbake are really sparse.

Anyone have any advice on how to get up to speed with it, short of
embarking on my own embedded linux project?

I also left long ago

[Apreche]

In college I was big into Gentoo. It had it's problems, sure, but when you got it working, it was terrific. Then after the college days were over, and I started working, I had a lot less free time. I realized that with Gentoo I spend a lot more time working on the computer itself as opposed to using the computer to do other things. I've switched to Ubuntu, and haven't looked back since.

Re:I also left long ago

[Tiger4]

"I've switched to Ubuntu, and haven't looked back since."

That is EXACTLY how it should be. I love the cool techie DIY of computers too. But ultimately it is a tool that is supposed to do something besides occupy my time. It needs to move out of development and into operations and production. The users, the people that can benefit from computer technology, don't want to build a computer or the OS behind it or maintain any of it, they want a magic wand that solves some problem for them. The maintenance is just a (small) price they are willing to pay to get the wand.

Re:I also left long ago

[r_jensen11]

I realized that with Gentoo I spend a lot more time working on the computer itself as opposed to using the computer to do other things. Funny, I underwent the same transformation. My solution, though, was to stick to packages listed in Stable instead of ~. The only times I've had problems with my computer in the past 6 months or so have been when I tried adding something cosmetic, like adding a bootsplash. I was able to avoid the problems many people faced when upgrading Gnome from 2.18 to 2.20 (or was it 2.16 to 2.18, whenever expat was changed) by following instructions.

Oh yeah, I've also had problems with getting compiz and the last two or three fglrx packages to work, but then again, that's cosmetic as well. So is trying to get the open-source ati drivers to work. Basically, it was when I realized that all I really use my computer for these days is playing music, watching videos, checking email, and checking websites like /. and the beeb, that I concluded that what I have had for a long time is 'good enough.' Now, the vast majority of my package upgrades are for security fixes, not for some new eyecandy.

But I was able to accomplish this even by sticking with Gentoo.

Re:I also left long ago

[Sancho]

I wish that I had the mod points for you. That sounds exactly like what I went through. In fact, I sometimes wonder what I might have made of myself if I'd spent all that Gentoo-tinker time in other ways.

Gentoo could be really spiffy, but the execution just isn't there.

I use FreeBSD for servers. It uses something similar to portage (I believe portage was even based upon it), and it's pretty damned stable and easy to administer. I do believe that they could learn something from the improvements that Gentoo made, as long as they actually got them right. For my workstations, though, I stick to Ubuntu, because it just works, and it doesn't require me to tinker.

Re:I also left long ago

[fast turtle]

One of the reasons I made the switch to Gentoo was the other distro's started to feel like MS with all the unneeded services starting at boot.

WTF is it with all of them? I prefer the Slackware setup. Nothing starts until the user decides what purpose the system has and Gentoo gives me the same level of control. I also get the ability to decide what dependencies are to be installed through the use-flags. For example, Do I want Imap support in all apps that have that option? Hell no, I don't need Imap support at all. Same for Ldap. Damn system is a single user desktop.

That's what I like about Gentoo. The ability to configure it for my usage and unlike most of the new Gentoo users, I have never performed daily/weekly updates to my system since I follow the "If it aint broke - Don't fix it" philosophy. This means the only time I update the system is when a security fix actually impacts one of my primary applications. Otherwise, I'll run things until they're so damn old people don't recognize them anylonger.

As to the problems with Portage, there is a project called Palidus to create a package manager written in C/C++ that can be easily fixed/updated/patched/extended by any distro instead of the mess of spaghetti that Portage has become.

Re:I also left long ago

[Breakfast Pants]

I guess you somehow managed to miss out on the libexpat fiasco?

Re:I also left long ago

[Anonymous Coward]

there is a project called Palidus to create a package manager written in C/C++

1) Learn to spell.
2) There's no such thing as "C/C++".

Re:I also left long ago

[r_jensen11]

I guess you somehow managed to miss out on the libexpat fiasco? Yep. All I did was follow instructions and didn't have any problems. I really couldn't gather why some people were having the problems they had. Maybe it's because was in the process before that time to thin down my packages.keywords

no it won't fail silently

[Laebshade]

The first line is redundant, and also by the first line being present, the second line contains an unnecessary command: "u". This is a switch to upgrade the package; since it was unmerged in the first line, it isn't necessary. It should be written as:

emerge -u gentoo-leaderships

- or if you want to see what version and flags will be in use when it's upgraded, try:

emerge -uav gentoo-leaderships

Even the original code won't faily silently; you can view the stdout if it has failed, you know.

awwww

[Anonymous Coward]

what did we accomplish at Microsoft that was worth leaving Gentoo for in the first place?

Duuuuuuuuurrrrrrrr!!!!!!!

DANGER! This guy is not to be trusted anymore

[Anonymous Coward]

He worked for Microsoft and now pushes for Microsoft based technologies(.net based), then he is not to be trusted at all. It's too dangerous for gentoo.
He was mandated by MS to sabotage from the inside the main source based GNU/Linux distro, since no proprietary software can compete with its flexibility and power thanks to the direct use of the source code. The future of open source is not binary package but source package where no proprietary software can go!
Moreover, thanks to its flexibility, gentoo gnome is not infested *by default* with toxic Microsoft based software like mono.
Beware, The wolf is trying to destroy the Little Red Riding Hood.

Re:DANGER! This guy is not to be trusted anymore

[Anonymous Coward]

Even a source based system can add a patch at the last minute or have an issue. Source code is altered all the time. Not to mention the fact that end users don't want to wait for source to compile endlessly. The best solution is to offer source and packages. Let people have choice.

Your claims about Microsoft are a bit over the top.

Re:DANGER! This guy is not to be trusted anymore

[Anonymous Coward]

gentoo does not need a president. if it did need one, they should consider other alternatives.

Re:DANGER! This guy is not to be trusted anymore

[Anonymous Coward]

Slashdot: Proving that even nerds smoke crack.

Choose one: fast or stable

[DrJimbo]

I've been using Gentoo for years and I love it. In the great FOSS tradition, Gentoo is all about choice (read: control). In the early years the stable branch was not very stable and the forums were filled with emerge problems in the stable branch. In the last few years those problems have almost all gone away. My system is mostly from the stable branch and I almost never run into a problem when I do updates. This is a very good thing[TM] IMO. I'm very glad the devs keep packages out of stable until they are almost always problem free.

But with Gentoo I can keyword-unmask specific packages if I want to be more on the bleeding edge and if I am really adventurous, I can hard-unmask packages. Yes, running a mixed system can require me to unmask dependencies but that has never been a nightmare (for me) and I don't see how the Gentoo devs could possibly prevent it. In fact, it should give you insight into why some packages seem to take longer than usual to make it into stable: all their dependencies need to go stable also.

It could well be that Gentoo is not the right distro for you. If your top priority is ease of use then one of the binary distros would probably be a better choice. But if you want the maximum control over your system, Gentoo is hard to beat.

How non-tech people can help

[Endymion]

Your post reminds me of some of the (very) old slashdot stories asking the question of how non-technical/non-programmers can help the F/OSS movements. The response usually came in the for of "there's lots of stuff like documentation (real writing, not tech writing), art, etc, that are also needed."

This was and is true, but I think another big category was forgotten: management. That for a good project to succeed, the techies should keep doing techie stuff, and be shielded from politics and dull aspects of business as much as possible.[*] This Gentoo problem is a prefect example of this. The techies don't want to do business, so that's an area that non-techie people could really help.

I hope the pull through with this, regardless of the solution - I would hate to have to switch to another distro...

[*] - See: Brokes and the MMM, where he talks about the "surgical team"

Great news

[wytcld]

I've been using Gentoo on production servers (and my desktop) since the first year it was out. It used to be a very solid project. It still has much better documentation than, say, Ubuntu. Originally the speed of the custom-compiled stuff was important to me, because of lower-end hardware of the time I had in use - which it did well on (once compiled, of course). The other main virtue, compared to Red Hat or Debian or Slackware of the time, was that it was easier to keep an up-to-date server running without having to do a fresh OS install every year or two.

But over the last year especially Gentoo has gone into steep decline. Upgrades of major stuff come with "upgrade guides" that leave out major things that commonly get broken. The Gentoo bugzilla is manned by kids who compete to close bugs while insulting the intelligence of anyone who'd dare file them. Older libraries which take little space and conflict with nothing are removed without choice or warning when newer packages are installed, and it's just tough if your production server has stuff installed doing useful work that depends on those libraries. Meanwhile the Ubuntu project has worked very hard to become the most-safely-upgradeable Linux (I'd imagine Red Hat must have improved too; but I hate rpms too much to want to try it again). And hardware is so fast now that for standard server stuff there's much less to gain from customized compilation.

For those who say that Gentoo is fine if you just keep a spare system to test upgrades on first, that's bull. Stuff will break on nearly-identical servers that are just slightly different in their versions - that is, going from 1.17 to 1.19 on a app may break, while going from 1.17 to 1.18 to 1.19 works fine. And the breakage can show up tangentially, not just where you'd most expect it. So you'd have to keep a test server for each production server, and very carefully keep it just one step ahead in sync. Plus you'd need to keep it under some sort of dummy load, since some breakage only becomes apparent in production, not in idle use. The real solution there would be for Gentoo to start being responsive to its bugzilla reports again, immediately fixing any breakage caused by new packages so that instead of letting hundreds or thousands of people trip over the same stone, the paths are kept free and clear.

If Robbins comes back, Gentoo could shine again. If he doesn't, it's about over.

Re:Great news

[wrygrin]

that's eloquently stated.

as ever, the amount of tangents in the slashdot discussion is both interesting and distressing. for the subject at hand, it would be interesting to hear about some authoritative responses from the gentoo officials, but i see nary a trace of that - no official responses in gentoo-devel or gentoo-user, and the single focused response included on planet gentoo makes no claim to authority (and would be more compelling if the logic or even grammar were more polished).

maybe it's too soon - but i sure am curious about how this is going to shape up. i'm not unhappy with gentoo - the bottom line, for me, is that it's presented fewer maintenance roadblocks than any of the distributions i've tried, over the years, including slackware, redhat, fedora, a little ubuntu, and a little less plain debian. turbulence and churn upgrading has felt difficult in the last year, however. i would be sad to lose gentoo - and i think that lack of effective, constructive response to the challenge that robbins has identified would mean losing it.

Re:Great news

[macshit]

The other main virtue, compared to Red Hat or Debian or Slackware of the time, was that it was easier to keep an up-to-date server running without having to do a fresh OS install every year or two.

Er, have you ever used Debian? The above is true of redhat, but it most certainly has never been true of Debian -- the Debian project puts vast effort into making sure upgrades always work (including code to cleanup problem situations), and it's essentially never necessary to reinstall a Debian system unless you do something really stupid (outside the packaging system).

Meanwhile the Ubuntu project has worked very hard to become the most-safely-upgradeable Linux

... mainly because they get the vast bulk of their system from Debian. The magic is Debian, not Ubuntu.

Re:Great news

[Enahs]

Upgrades of major stuff come with "upgrade guides" that leave out major things that commonly get broken.

Oh, wow. That was one of the reasons I decided to stop using Gentoo...three years ago. Sorry to hear that it's getting worse.

Really, if their leadership was worth a hoot, there'd be a primary emphasis on stability and upgradability, and there'd be a next-generation Portage by now. Instead, oh by gosh by golly, there still seems to be infighting, and the rule of the day seems to be that Gentoo ricers are all dern impressed that they all use Gentoo.

At this rate, it'll die off, and if there's no progress, that'll be fine if regrettable.

Re:Great news

[Anonymous Coward]

Testing new packages on a testing server is bull? Well, that's a big bull... trying out new packages rolled out on production with no testing...
And just mount /usr/portage from the testing machine and let the binaries roll on the production once you think it's working... it applies to any distro... that you normally test a damn before you roll on live...

Re:Great news

[janiskracht]

> If Robbins comes back, Gentoo could shine again
This is it exactly.. I've been running Gentoo for at least 6 years.. I'm afraid now to update the simplest packages because so many packages are generally broken.. My husband has abandoned gentoo at least a year ago, for his internationally marketed software on his development systems (he started using it shortly after I did..) I'm looking forward to Daniel Robbins being involved once again so that some sanity will return here... I love Gentoo. It's been said that one cannot run a project by committee... and I believe that is true unless you have ONE person who is willing to take care of details.. someone with a vision I think Daniel is it...
Janis
www.filegate.net

Re:Great news

[Anonymous Coward]

"Originally the speed of the custom-compiled stuff was important to me, because of lower-end hardware of the time I had in use - which it did well on (once compiled, of course). "

          Funnily enough, I have put ubuntu onto systems as low as a PII 333, directly over a gentoo install, and the ubuntu install actually ran a little faster. (And, of course, forget about the install times -- they are DEFINITELY better with ubuntu 8-). This certainly isn't logical, the custom-compiled apps should be faster... but for some reason they are not.

"The Gentoo bugzilla is manned by kids who compete to close bugs while insulting the intelligence of anyone who'd dare file them. Older libraries which take little space and conflict with nothing are removed without choice or warning when newer packages are installed, and it's just tough if your production server has stuff installed doing useful work that depends on those libraries"

          These are my problems at present. I still have some gentoo systems, because they work and I don't feel like backing everything up on them and "starting from scratch" as it were.. but my others now run Ubuntu. When I've reported bugs, they actually get fixed instead of just closed 8-). One of my remaining gentoo boxes in fact runs several items that were simply pulled from portage.

          I must say ubuntu is a damn nice desktop.. to replicate lots of USE flag tweaking etc. I had on my gentoo install (and then some -- the automount, automatically handling printers, ipods, etc is nicer under ubuntu..) I literally installed ubuntu, installed ubuntu-restricted-extras (codecs, flash, java, etc in one shot), installed mplayer (totem works but I like mplayer..), installed mythtv-frontend and pointed it at my myth box, set it to share via nfs (it brought up the pacakge manager to install nfs-server pacakge), and set a few entries in my hosts file and a few remote nfs mounts in /etc/fstab. I love the whole idea of USE flags, but ubuntu seems to pick the "right" defaults almost without fail anyway 8-).

"For those who say that Gentoo is fine if you just keep a spare system to test upgrades on first, that's bull. Stuff will break on nearly-identical servers that are just slightly different in their versions - that is, going from 1.17 to 1.19 on a app may break, while going from 1.17 to 1.18 to 1.19 works fine. And the breakage can show up tangentially, not just where you'd most expect it."

          Yes! I have a few systems that I sure *thought* were setup identically, and as you say, some package will behave differently between them. An odd one I have now, qemu shows "Could not configure '/dev/rtc' to have a 1024 Hz timer." on one system, not on another.. why? I don't know.

Classic Gentoo?

[TaoPhoenix]

I heard a rumor once that Gentoo was not for the faint of heart incoming Linux converts. If Parent Poster's remark is on target, I think I just found out why.

I sortof understand Debian's philosophy as the Ultra-Scrubbed distro which includes things like Icy Cats instead of Firey Foxes, because the animals don't all play nice together on a branding level. (Will someone ever port Icecat to Windows, or is that a contridiction in terms?)

What exactly is Gentoo's theme? Other than being super-componentized, is this kind of administrative fallout systemic to the distro?

Re:Marking package/version as unstable

[ion.simon.c]

What's wrong with:
# flagedit =package-cat/package-version -- '~x86'

No muss, no fuss, and you only have to do it once?

Also, --depclean *crashes*??? Can you post a backtrace and any error messages?

Re:Why should we care about compiler flags?

[ion.simon.c]

To me, Gentoo's not about the crazy flags you can throw to GCC. It's about the flexibility of portage. Ex: can you get debugging symbols for *everything* in your binary-only distro? Probably not.
WRT compiler flags: I've gotten along quite nicely with 'CFLAGS="-O2 -mtune=${CPU} -fomit-frame-pointer -pipe"'.

Also, your "Archetypal Gentoo Builds" idea is good. Perhaps drobbins will work towards something like this in the future?

Tagging this is like hitting slow pitch softball

[siglercm]

Recommended tags:

gentooisdyingnetcraftconfirmsit

robbinstannedrestedandready

Please reply, adding more (hilarity) as needed ;)

Reasons for leaving Gentoo seem exaggerated...

[Brane2]

I'm long time gentoo user ( I've started around version 1.4) and am more or less stisfied with it. I mean, it definitely has its flaws, but I haven't been able to find substitution that could "scratch my itch".

I have tried Ubuntu as everyone around me were advocating it, but found that while it has much prettier installer and things tend to work out of the box, deep down it's actually inferior to Gentoo.
When things work smoothly in Ubuntu, everyone is quick to point out those maybe few minutes and a CLI command that Ubuntu has spared you, but no one mentions those cases when things don't work.

Each distro has its framework which combines many pieces of open source mosaic, but things get interesting when some piece in mosaic develops a flaw that is not immediately obvious or it affects some portion of users. I don't care for a few seconds spared during installation nearly as much I care for infrastructure support in cases that don't work.

WRT to Gentoo's imminent death:

1. If its going to happen, it won't be soon.

2. All problems of Gentoo can be traced to its origins. At the time, its creator found his pleasure in homebrew approach and wanted to have something that works in some way much rather than trying to get it right first time and also answer many organisational, commercial and law questions.

So now we have Gentoo Organisation, Infrastructure and Distro in the state of Russian Orbital station MIR jsut before its death: there are many interleaving and intervening systems with many semi-documented patches and changes and whole shebang is far from original specs. I mean, evolution is a ni ce thing, but it has its limits. When it reaches its limits, maybe its time to use accumulated knowledge and experience to make something new...

3. WRT to Drobbins, I don't know the guy personally and have nothing against him, but I'm not sure that having him back is a good idea.
He had the chance but has proven unable to make Gentoo his life, so now he's coming back, faced with similar problems ( needing money for RL but being strawn between his hobby and bussiness) and unable to learn from his mistakes and use radically different solution this time.

4. New Gentoo should start from scratch with its policy, organisation and web/distro infrastructure while good old Gentoo I is living on...

My reason was: couldn't build the core of it with

[AntrygRevok.net]

My reason was: couldn't build the core of it with package self-tests enabled.

It was core-utils or linux-utils that couldn't compile when checks/tests were enabled,
and if even a single package compiles but is scrambled,
I don't want it near any system I admin.

Why can't they have checks/tests enabled as a final auto-check before marking something as OK?
( and no, I wasn't using the "~" stuff )

MOD PARENT UP!!!!

[Anonymous Coward]

nt

spot on!

[FreeUser]

I'm sorry, but that is total crap. I have been using Gentoo on production servers which I *do* keep current using stable (not bleeding-edge) packages. This is a large shop with many servers. I have never looked back since switching to Gentoo. Everyone who moans about emerges failing and having to run revdep-rebuild often must be doing something wrong. I've had to run revdep-rebuild once when I upgraded libexpat. So what? It took like 2 minutes.

Absolutely right. We deployed Gentoo at my former employer (a Chicago Hedge Fund trading in most major markets) as servers, trading desktops, and developer machines, and it proved to be ideal. Most people don't seem to realize that Gentoo is a meta distribution, allowing finely grained control over exactly what is included (down to the version and compile-time options). Manage your local portage tree, mirrors, and define your site-wide default USE flags, and you have a tweaked platform that is exactly right to your needs. Taking the time to do this has loads of benefits--and ultimately takes less time than trying to shoehorn Fedora or Centos into doing a job it isn't optimized for (and sometimes doesn't support at all).

Anyone unable to make Gentoo work in an enterprise environment doesn't know Gentoo. Now, it isn't for everyone (my current employer uses CentOS, which is fine for them, but wouldn't have worked for my former employer at all), but Gentoo is far more useful in far wider circumstances than its nay-sayers would have us believe.

*OT*

[Anonymous Coward]

not official firefox but you might get interested in http://www.getswiftfox.com/deb.htm this guy keep fairly update it (if you haven't heard it before of course )

: For example, I still didn't find any place that offers a .deb of the new Firefox Beta 3. Anyone willing to point me to one?

Not for everyone

[seibu]

I see a lot of people who have this or that frustration with Gentoo, but in each case it's pretty easy to see what tool or approach they overlooked which could easily have solved their 'problem'. Gentoo requites some effort and knowledge to use it properly, but it's hard to imagine how the distro could avoid this while still offering the same immense flexibility.

Of course somebody is going to prefer Ubuntu if they have fairly mainstream needs, and dislike spending any time configuring their system. But tell me what other distro could simultaneously drive an arcade machine with 15Khz monitor, a tweaked real time kernel for pro audio and a full suite of bleeding edge (and I mean *bleeding*) source-compiled soft synths and sequencers, all running flawlessly on a Pentium 3?

Useful

[bill_mcgonigle]

The only way I can think of to produce a totally generic solution is to use software that already exists for calculating optimal settings and for profiling packages, and produce as comprehensive a table as possible of what permutations work for what systems.

You might want to expand on that if you have some time and the energy to write an essay about it. Maybe a tool that uses QEMU or xen while being nice'd +19 for us to run on our machines with spare capacity. Bandwidth might be the highest cost.