Slashdot Mirror
Debian GNU/Linux 4.0 Set for December
dolson writes sends in a heartening update straight from the Debian project's news page: "The Debian project confirms December 2006 as the date for the next release of its distribution which will be named Debian GNU/Linux 4.0 alias 'etch'. This will be the first official release to include the AMD64 architecture. The distribution will be released synchronously for 11 architectures in total.
At this stage, the upcoming release will ship with Linux 2.6.17 as its default kernel. This kernel will be used across all architectures and on the installer. A later version may be selected during a review in October.
New features of this release include the GNU Compiler Collection 4.1 as default compiler. X.Org will replace XFree86 as implementation of the X Window System X11. Secure APT will add extra security by easily supporting strong cryptography and digital signatures to validate downloaded packages."
OK, I know that the various testing levels have had updated software for a while, but pushing this volume of changes to the mainstream distribution does seem like something of a shock. Debian's historic reliance on "tried and true" versions seems to be giving way (at least partially) to the realization that many people don't want to use it because it lacks significant feature updates.
I'm impressed.
You can never go home again... but I guess you can shop there.
You know, funny you should mention that. I've been using Ubuntu for at least a year, first tried it about 2 years ago. When Dapper came out I was amazed, all the configuration tools, menu editor, update manager and such.
I'd used Debian before, but not a lot, probably around GNOME 2.6 and lower. That all certainly wasn't there. Then, I decided to fire up VMWare and install Debian Etch just to see how things are moving. It was practically Ubuntu without the splash screen and Add/Remove Programs in the Applications menu.
Now, granted, I know that is certainly not the only thing the folks at Ubuntu have been up to, but it goes to show that Ubuntu isn't the only one making progress in the Linux world. Debian is still chugging along, faster than ever it looks to me, and Ubuntu is benefitting from that more than anyone.
I found that 2.6.17, with the improved IO handlers, definately added a performance boost to my machines. The main headaches I've had with testing have revolved around X.org 7.x being quite a bit different from previous versions (more componentized) and issues with getting it to work with the NVidia stub (you need to tell it where to find the new lib location), etc.
However, all-in-all I've found that running Debian/testing has gone pretty well, and Debian/stable+backports has worked pretty well too. I'll be looking forward to when the features in testing happily merge back into stable.
Oh, and hopefully the rather-cool FPS Nexuiz will merge into stable as well, as it's pretty impressive to see something like that ending up open-source and available in the standard repositories (it's available in testing+ right now). It's also the first OSS app that's really given my graphics card a run for its money.
The long testing part is due to the very big collection of thirdparty packages that Debian has, along with very liberal rules for package dependencies back and forward through various releases. I'm sure that is a major headache for the maintainers except for the "was-a-maintainer" that have left for something else.
So, long testing period does not imply higher quality with respect to other Linux distributions, or *BSD for that matter.
Official AMD64 support is a big one. The new kernel may-or-may not be necessary, depending on whether they backported the fixes for AMD64 to 2.6.8 or whatever they're stuck on currently. That's also important, because the bugs were pretty severe (random thread crashing.) SecureAPT is a nice touch as well. Xorg is a reasonable and probably not-too-disruptive change. GCC could be disruptive, but probably won't be that big a deal.
Flamebait, pure and simple. Listen up, Debianistas: the only hard and fast requirements are encoded in the licenses under which software is released. So, comrades, point to me how the ubuntu project (or any of the other Debian daughters, like, say Xandros) violate the terms of the licenses under which software in the Debian project is released?
How about "Not at all?" Take your bitterness, compact it into a pill, and swallow that.
The whole Debian/Ubuntu internecine bitchfest reminds me a lot of the communists I knew on campus--the Maoist faction couldn't even be seen with the Stalinist faction. Did anybody but them care? No. But I'm sure they had a lot of fun in their respective cell meetings, counting the meagre takings of their pamphlet sales (which had to be on alternate days, lest they have to share space with the traitors from the other side)
My Linux install started life as Slackware 10.2 (2.4 kernel), but it sure doesn't look like it anymore.
"I've spent my whole life figuring out crazy ways to do things. It'll work." -- Montgomery Scott, "Relics"
Apache 2.0 has been in Debian for ages. I'm afraid that you, sir, are a troll.
Also, the plans for 2.2 migration speak about having 2.2 as a replacement instead of putting it side-to-side, so no, Etch probably won't have that Apache 2.0 you want.
MySQL 5.0 was released in October 2005, Sarge in June 2005. So...?
The creatures outside looked from Alt-Right to Antifa; but already it was impossible to say which was which.
MORE TIME TESTING = MORE STABLE PRODUCT
:)
Or an outdated product, as I've come to realize. I have a rented server running Debian, which has given me nothing but headaches because some of the packages are horribly outdated, namely PHP5. I mean, PHP 5 is what, over two years old now, how come they didn't think it was stable nor tested? This is one of the reasons why the next server I rent will be running something else. Better to have "untested" (use that word carefully) stuff working than no stuff at all. Must be that "security-through-obsolence" paradigm rearing its head
Lousy analogy. Debian and Ubuntu are not on the equal footing - one's dependent on the other. Unless your Maoists are dependent on Stalinists, or vice-versa. It is in the interest of Ubuntu and other Debian-derived distro to ensure Debian remains healthy.
If Ubuntu has violated no license or broken no law, then would people do us the great favour of shutting the fuck up already? Goodwill is not enforceable. It may be desireable, but it is not a necessary condition for progress in the Free Software ecosystem. Is it just me, or is the Debian project moving a LOT faster now that the Debian Daughter Distributions--Ubuntu included!--have vastly expanded the pool of developers, testers and users?
We don't have to like each other. But it would be nice if we could appreciate what we mean to each other collectively.
Or an outdated product, as I've come to realize. I have a rented server running Debian, which has given me nothing but headaches because some of the packages are horribly outdated
I sure wouldn't want to have a debian stable server where I couldn't add a few repositories or pin some packages from testing. I like the stability of the groundwork, but you have to put some flexibility into it yourself. Debian stable is frozen in time, I'd say 18 month release cycle plus 6 months because they don't put bleeding edge in there to begin with, so being two years behind is by design. Honestly I wish I could jump between debian stable and ubuntu LTS... 9 month release cycle with long support on every version if I choose to hold back would be outstanding.
Live today, because you never know what tomorrow brings
Probably you're ack'ing our work on desktop environment task, a subset of the work in the tasksel package. Everybody else can benefit of this just selecting after base installation that they want the desktop environment! I'm directly involved on this and added update-manager and other stuff there. Feedback is welcome. :)
If you're using 'testing' you can do aptitude update && aptitude install desktop gnome-desktop . Enjoy! More documentation to come with a new desktop related web page soon.
-- stratus
I find that unless you are building mission critical process control systems that need to be extremely bug free, you are better off using the Debian testing version than the official release, particularly if you have newer hardware that you want to be able to use.
You can do this if you like, I guess, although I would feel a little uncomfortable: My rule is "stable whenever it matters to someone else". I use testing on my own machines, but I've definitely found myself in situations where testing was broken (usually just due to large upgrades like libc6 or something, but still, more broken than I wanted to deal with) -- or, if the whole archive isn't broken, you can still get upgrades forced on you that change the behavior of the system in unpredictable ways and make you unhappy. Generally, the increased stability of "stable" is worth it to me and my users when I'm doing any sort of administration.
It's useful to note that in the uncommon-but-not-rare case where you or a user wants a package upgrade from testing or later, you can very easily use apt to pull down the source and build-dep, compile it for your system and install it as a package with very little hassle. Do this for the packages where it matters, and you have a mostly stable system with the features you need.
At least when compared to MS. Three Debian releases between XP and Vista. And people say Debian is a slow mover.
Well, at least assuming that both "Etch" and Vista will hold their target dates... Is this a too bold assumption to make? Perhaps.
while true; do eject; eject -t; done
Unfortunately, stable and untesting are just terms that Debian uses to refer to the different releases. Debian testing is by far more stable than any version of Windows I have ever used, and for all pratical purposes it is Stable. They could just as well have labeled the Debian versions "new", "stable" and "old" than "unstable","testing" and "stable".
I'm an American. I love this country and the freedoms that we used to have.
You said yourself in your own post, that you used an old debian with GNOME 2.6, and now you're seeing new features in the new debian and you're impressed with the progress debian is making.
Now, I'm not arguing that debian isn't making progress, but... oooooooh, they packaged the latest GNOME. Any distro that happens to package the latest GNOME also has made all the exact same progress you speak of. So that's kind of meaningless. GNOME has made leaps and bounds of progress in terms of usability, UI consistency, things like that... credit where credit is due man, this is GNOME making progress, not debian.
Debian/BeOS and Debian/Plan9 would work just fine. You'd just have to accept that you were using an operating system that's not Unix, and not Not Unix either.
Creating a distribution of BeOS or Plan9 with the Debian mindset and many similar tools is certainly possible. However, in order to provide the true experience of either OS you'd need a dedicated team willing to write native programs in the style of those OSes and make sure that improvements made to the Debian/GNU Linux versions were reflected in their separate programs. The same thing is true of Debian/Win32 (except that they can't actually distribute Windows; perhaps Debian/ReactOS). Completely possible, but it's a lot more work than just porting things.
It's just a matter of priorities, really. There's a balance to be struck between having the newest stuff and having a stable distribution. The stable branch of Debian just balances its priorities very strongly towards stability. It's up to the user to decide whether that's what they want from their operating system. If not, there is other branches of Debian, and other distributions entirely, which can be used. Allowing wild variation in philosophy like this is one of the redeeming features of the "distro soup" that exists.
Why not just grab the latest v5 source from www.php.net and compile it yourself? That way you get the latest version and you control exactly what features you get (e.g. PEAR support / CLI support / PDFlib / libxml etc.)
Just because there isn't a package for something doesn't mean you can't have it. In fact, for things like php, apache, mysql, ftpd - it is almost always better to compile yourself from source. This way you maximise the apps efficiency and arguably enhance security as you disable any feature you don't need/want.