I for one hope that they don't come to an agreement. That way we'll see Mozilla on a lot of desktops. Even if they were to include an crappy release like Netscape 6, they're AOL users, they wouldn't be able to tell the difference. Right?
Seriously though, it would be nice to see Mozilla get some marketspace this way and it is getting good enough (especially speed-wise on Windows, Linux needs tuning) that most people would be happy with it. XP is not going to be running on your P166, so the requirements are going to be fine and I doubt most AOL users upgrade all that often. From what I've heard from some Mozilla developers, the AOL folks hate Microsoft with passion, so don't be suprised if they stick with their horse rather than ally with the evil empire.
Yes, I think I confused what I thought the parent was saying with Microsoft's stance. My bad, my post was a bit OT.
I agree with Ballmer saying that government funded work should be public domain.
I still think my comments are quite relevant to the issue though. Microsoft has been claiming that the GPL destroys IP, I was trying to refute that (and perhaps it belongs in a different thread).
You're exactly right. Microsoft seems to believe that the Linux community has some obligation to make the source code available to them for easy integration.
This is simply irrational. Linux is a competitor to Microsoft, as they have stated countless times. Does Microsoft complain that IBM doesn't make the source of (current) versions of OS/2 avaible or Apple the proprietary bits. No, because they are twisting the facts.
The GPL acts as a sort of protective measure, keeping undesired competition out. If you want to use GPL code, you have to open your code base. This specifically protects intellectual property because Microsoft (and others) can't grab it wholesale and use it in their products without making their source available. Doing so would put them on even footing with the open source competition, in that the open source/free software community would also have access to their I.P. and be able to incorporate Microsoft code.
Imagine if SGI had made XFS available under a different license, ala the BSD or X11 license. Sun, HP, and IBM would be able to port the source code to their version of unix and as a result, have a better offering. But the GPL adds an additional barrier to entry, they would have to do something they have no desire to do, make their code base entirely open and free. So, they will not use the XFS code and SGI's I.P. is being used in the way they want it - that is, they will ahve access to the source of any product using XFS. That is precisely what they want, because they want to be the guys selling you hardware that uses XFS.
They are in business here and open source makes sense for business reasons, not some altruist desire.
The Progeny team is participating in the Open Packages effort. We may end up seeing ports in Linux as well.
Re:Why i'm still not switching...
on
Mozilla 0.9 Out
·
· Score: 1
DJB absolutely does not want any changes made to the directory structure. That is the principal reason he gives for being so anal about distribution. The packagers WANT to change the directory layout - it provides additional support for them. Having a package install in exactly the same place everywhere makes it easy to admin everywhere.
I wholeheartly agree with this. Having everything in the/var/qmail directory is great. Why not make the license something like the BSD license with the explicit condition that you never, ever change the directory structure and if you make changes, change the name. That way it can be the default MTA.
The rationale is that there are frequency upgrades to documentation, and the website will always be up to date.
That's fair but ideally your documentation should reflect the current version. Your interfaces aren't going to change after they're written and the tools will still accept the same switches as when they're released. The documentation should be functionally complete for the version that's being released.
OK. Bernstein is a little quirky. It often comes with being an academic. However, he has done a lot for free software even in just considering his cryptography court battle. And he has written a lot of software as free as QT 1.0 if not free-er.
http://cr.yp.to/software.html
I think it comes with being DJB:) There are plenty of other academics who are a bit less unusual. He's put out lots of great software, this I will not dispute. It's rather interesting to see that he's able to write what is generally faster, more secure software as one person working alone than teams of developers.
In a perfect world, it would be nice if his license loosened up a little. But he who writes the software gets to choose the license, and DJB makes things as free as his own sense of what is right in software allows. If he were to GPL his
programs, then all the distributors would alter directory structures. And he thinks that is bad. So he allows unrestricted binary distribution with the exception that the distribution has to occur EXACTLY as the tarball build would make it.
Not quite that simple, from http://cr.yp.to/qmail/dist.html:
(3) the package's creator warrants that he has made a good-faith attempt to ensure that the package behaves
correctly. It is not acceptable to have qmail working differently on different machines; any variation is a bug. If there's something about a
system (compiler, libraries, kernel, hardware, whatever) that changes qmail's behavior, then that platform is not supported, and you are
not permitted to distribute binaries.
To me, all patches change behavior. In order to include qmail in the src tree of the bsds, you will have to make modifications to the makefiles, etc. However, he does specify that he may make exceptions if you email him, I may just have to ask him about it.
I'm not suggesting he GPL his software. Just make it a bit easier to include as the default MTA in linux and bsd (e.g. allowed _labelled_ changes). Though, I can see some people rejecting it solely on the basis of the directory layout...
Anyway, I'm beginning to see that I'm getting a bit anal about this. Just liking free software, I'd like his stuff be a bit more free (and send sendmail to the retirement home).
Offtopic, but for anybody having problems with junkbuster breaking certain pages with mozilla, you can fix this by disabling keep-alive in the Debug -> Networking section
Re:Why i'm still not switching...
on
Mozilla 0.9 Out
·
· Score: 1
djb writes great code, but his license is nuts. I love qmail and have thought about it a bit and have come up with the following legitimate problems with not allowing modified redistribution.
1. You will never see qmail included as the default MTA in any free OS. Whoever is doing the packaging/migration into the distribution is going to want to make patches at some point or perhaps change the directory layout. Not to mention that it interferes with certain goals of some projects to provide completely free as in speech software.
I can understand his qualms about insecure versions being produced. Why not address this in the same way as the artistic license by making it so that only djb can produce binaries called qmail and other people would have to change the name to indicate that they are distributing modified versions (call them OpenBSD-qmail or something).
2. I can't use his libraries for my own projects. One point that he raises on his web page is that he's got a much more efficient dns resolver than the one that ships in BIND. Great, he gets to use that in qmail and his programs, but if I want to use it, I have to ensure the user has a copy of djbdns installed just so that I can make use of a small part of the program. Is linking against his libraries even allowed under his license?
The other thing that really peaves me is that he ships what I consider incomplete versions. djbdns does not include installation instructions or man pages in the tarball (the latest from his webpage). Instead, he expects you to read them off his website. This is damn inconvenient and it suggests to me that he wants to measure the usage of djbdns or some such. If so, it's a pathetic means of doing it. I hate control of information, especially in a case like this. You can man-ify the html, but why make people do extra work.
I've also heard that he doesn't comment his code. Given that he doesn't mind people writing patches, why not make it a bit easier to understand what's going on.
qmail is great and from what little experience I had with it djbdns isn't bad either. I just wish he would lighten up a bit. At least some outside patches are finding there way into djbdns but qmail could use a minor update to clarify the documentation (drop support for inetd) and include 1 or 2 of the performance patches out there.
As somebody on slashdot previously wrote, "nobody codes more paranoid than djb." Hopefully qmail 2.0 will include some cool stuff we he has it finished.
Re:Half the ram and twice as fast?
on
Mozilla 0.9 Out
·
· Score: 2
The things I've read said that currently the Mac port is the worst of the three main ones in terms of stability/performance (not a dis on Macs, but just the way Mozilla is). However, I'm suprised to hear its that bad.
BTW, the interface is pretty much fixed. You can add and remove some of the buttons, but if you want real change get new themes. New themes are available thru the view, apply themes, get new themes menu (which takes you to x.themes.org's theme site). The Lopburi flat theme is great, if the guy ever gets around to removing the text labels from the buttons.
BSDi is considerably different from FreeBSD. They both started around the same time if you count the roots of FreeBSD in 386BSD but have been developed separately.
While BSDi has borrowed some code from FreeBSD, their trees are divergent enough that when FreeBSD began work on SMPng, they had to code from the ground up (sharing some common primitives) because the BSDi code wouldn't port over to the FreeBSD kernel.
Although I am not particularly familiar with BSDi, I can assure you that they have their own developers to implement features on their own when they don't find something appropriate in Free|Open|NetBSD.
Not to mention you have your facts on Sun wrong too. They did use BSD code to create the first few versions of SunOS (pre 5.x) but at the time, the code was not free. You had to pay for a license in order to get a copy. They may have even had to pay an additional fee to produce a derived version for profit.
You have some good points, but your facts are lacking. Get a clue, fool.
If you want a window manager that's truly stripped down but still has plenty of functionality, try pwm or Ion which both make good use of the keyboard and run in about a meg of ram.
pwm somewhat resembles BlackBox without the bar at the bottom or iconify, adding the ability to attach windows together into frames for each navigation. You can move thru the frames with hotkeys, keeping the desktop more organized.
Ion takes it a bit further and forces all windows to maximize to fit a fix sized box (frame). You can have multiple programs in a given frame and can navigate thru them or move from frame to frame with mouse or keystrokes. It's truly odd and a big departure from traditional CDE or Windows knockoffs.
These are where real innovation and progress in window managing is. Making something that doesn't just look pretty or makes Windows users at home, but who would have guessed, actually help you manage windows better.
I'm not surprised at all. They'll do whatever they think they can get away with that will help them make money.
They're undercutting us by paying for Verizon the ISP's losses with Verizon the telco's profits. Since the ISP and telco are actually two separate companies on paper (Verizon Communications is a parent company of Verizon Internet Services), AFAIK they aren't supposed to share money back and forth. I have little legal knowledge, so I could be totally wrong on this and if I am my whole argument is meritless.
If I am correct, I have no way of knowing whether they are doing that, but if they aren't Verizon Internet can't be making any money. Bandwidth/admins/other costs aren't cheap enough they are able to make it on $7.50 net income. There's no way they could afford to do that for more than a few months.
I'm not suggesting they should offer us some advantage, but rather play by essentially the same rules that we have to. They charge everyone $32.50 as a "line charge", fine, then expect everyone to compete on the ISP charge portion. Don't send have the telco Verizon Internet a check every month to make up for losses.
One of the primary problems is that there is a flat rate for line charges. If Congress removed the tariff that dictates the $32.50 price, then broadband might be a lot cheaper right now. (Under the terms of the tariff, the price for ISPs with over 500,000 subscribers goes down - so it is possible that Verizon Internet is doing this, it's just damn hard to figure out what's going on).
Yeah, I don't see what any real advantages Verizon has there. We outsource our news to Supernews and I haven't heard any complaints from customers. From my experience, it's not bad, pretty reliable when I've used it. However, I haven't done many binary downloads, I believe our cap is somewhere around 256kbits so it may not be what you need.
One main advantage that they have is 24 hour tech support. Right now, we don't, though we have considered it.
As far as the other services go, they're all pretty inexpensive so we do them inhouse. It's really not that hard to keep a system running if you know what you're doing. A couple FreeBSD servers running qmail and a Windows NT machine from before I got there. We have our outages when something breaks but generally things are pretty good.
One thing I like about our setup is we use NAT, which no one else seems to do. It's rather nice since we can allow people to have multiple computers connected without using lots of public ips. Of course, NAT has it's own disadvantages but we charge $5 per static ip so it's not too bad.
If you want, I'll email you our url (who knows, you may have already tried us). I not posting my email address or our info because I prefer not to find somebody at Verizon getting ticked off at my comments.
I work as a sysadmin for a small isp in SoCal and one thing that's makes it really hard for us to get by is some of the ways in which GTE/Verizon and PacBell try and squeeze out the little guys.
For example, Verizon is dictated by law that they must charge $32.50 for the DSL line (not including ISP service). That was fine when they charged a total of $49.50 for a complete package. But in order to remove competitors, they have lowered their prices to $39.99, causing their ISP division to take a loss presumably, while they pay for it with profits from the telephone division. This makes quite hard for ISPs using Verizon to supply the physical link to make it on such a small margin.
PacBell used these tactics as well and has virtually eliminated all the competition in their areas. They set their price at $39.95 and once they had most of the market, they raised it back to $49.95. This only works when they have large cash reserves or can abuse their relationship with the telco to fund price cutting with the DSL line profits
Traditionally, Verizon offered free modems to all customers who signed up for a 1 year contract with any ISP. They've since stopped that offer as well and made it so that only their customers get the modem, while ours must pay $200. Gee, which company will the customers sign up with.
This sort of crap makes it really hard to make a profit or even stay in business. We've managed to stay alive, but a lot of other ISPs in our area have gone under. From what I've heard Verizon doesn't have particularly good service, they just win through anti-competive behavior and ISPs can't afford to sue and don't have the market share to make them compromise. Definitely bad for the consumer and us as well. Hopefully, Verizon will decide they have enough of the market and raise their prices back, allowing us to make a bit more money.
For one thing, OpenGL puts most of the code in the driver. That means that the vendor has to write a good bit of the driver whereas with Direct3D you use Microsoft's implementation and hook it into your hardware.
This has the nice benefit of making new drivers pretty fast right off the bat (see ATi's troubles getting decent RagePro GL drivers), but everybody's gotten mature enough OpenGL drivers that this doesn't matter much any more. Not to mention with the consolidation of the graphics market, no one will be writing new GL drivers, just revising their existing versions.
I had the same problem with the mouse and got around it by not using moused. Instead, just use the raw device in X and set your device to/dev/psm0 (or whatever) and Protocol to Auto.
AFAIK, this was fixed by 4.2 anyway. I don't use moused as I find little need the mouse in the console. Terminals in X or straight keyboard for the servers suits me just fine.
>Meanwhile, FreeBSD suffers from old, stodgy code written by core Unix developers with >1980's-style development practices. Sorry, but the art and science of software >engineering have improved vastly, and Linux developers are *NOT* afraid of saying, >"Okay, our implementation of xyz is flawed, let's throw it away and do it again". So >Linuxleapfrogs FreeBSD and will continue to do so.
Uh, ok, let me see if I can feed the troll here. I can count some of the things that have experienced major retrofits in the recent past:
3.0 series - implementation of SMP (not so good)
port to the Alpha architecture
complete overhaul of SCSI layer to move to cam (ftp.freesoftware.com would be nowhere near as good without this)
introduction of vinum software/hardware RAID
4.0 series - major rewrite of virtual memory by Matt Dillon (not the movie star, and I believe his work was helpful to reworking linux vm as well)
IPV6 support - all new code written by the
USB support, stable a few months before linux
move to pcm audio drivers as the default drivers for sound cards, replacing the old OSS code (which still is available for compatibility)
integration of IPfilter as an alternative to ipfw
implementation of pthreads - licq and friends wouldn't even compile in the pre 3.4 era
DRI kernel modules - can be tough to get working (voodoo 3+ w/ xf 4), but it's there if you've got the time
5.0 series - SMPng in progress (should put FreeBSD back in competitive smp territory on a par with linux 2.4 hopefully)
libh project to modularize the install process (allow X and text installers) and break the system into smaller packages
openpackages - use the same build scripts for all BSDs (not really 5.0 but still a work in progress)
ongoing work on porting FreeBSD to sparc, powerpc, and arm
kqueues - which look like a good alternative to poll() and select(), though I have no experience with them
So I don't know what the heck you're talking about. Sure, Linux tends to move things into the stable branch more quickly than BSD and there are a lot more releases, but we get along just fine. From what I've heard, the FreeBSD kernel is a bit more elegant than the Linux kernel because code is less likely to be included unless it meets the committers' standards.
Linux development moves at a breakneck pace, FreeBSD is more conservative. What does that tell us? Absolutely nothing.
This is a bit of a flame, but some of the reasons that there haven't been as drastic of changes in BSD as Linux is that some systems have been high quality from the get-go. They drew on well-tested mature code (BSD Lite) that had already had a lot of the bugs worked out.
How many times has the Linux TCP/IP stack been reworked, like 4? The only place where I can see a halfway coherent argument is in the filesystem. It would be nice to see a journaling filesystem that would give better metadata performance than FFS + softupdates, but all the implementations are under the GPL. As it is FFS does fairly well and is more reboot tolerant than ext2 and until recently had Linux beat for large (> 2gig) files.
FreeBSD puts out a quality, coherent system that is stable and performs quite well on uniprocessor systems (and soon SMP). So long as they keep doing that, I'll be a devoted FreeBSD (and Debian) user.
Re:Good question in "talkbacks"
on
QT Mozilla Port
·
· Score: 3
It appears that the real problem may be with how gcc and mozilla are interacting. Please see this status update, in which it states that gcc is generating code that is twice as large as what Visual C++ is generating. If Linux builds are executing twice as many instructions, of course the Windows builds will be faster. Further follow-ups don't show much progress other than that switching from egcs to gcc 2.95.2 reduced size by about 5%.
Unfortunately, using gcc 2.95.2 requires a libc upgrade. I just installed the gcc 3.0 3/20/01 snapshot and am going to try a build tonight to see if it makes much difference.
Since pavlov was moved to help out with libpr0n, I haven't been seeing many updates about general Linux performance. Back in November thru early January, this was one of the main things the Footprint team was working on. pavlov seemed to be one of the most knowledgeable guys in the area and there were several people assigned to it. It seems they all got pulled elsewhere.
To really make Mozilla work well in unix, they need to put the focus back on speed and also take a look at the memory leaks that plague the linux version. After browsing for a few hours, Mozilla tends to swell up from around 20 megs of ram to 50 or 60. They had some graphs profiling this problem, but those seem to have stopped being updated as well.
On the whole, I find Mozilla pretty darn useable at 0.8.1. I run it on Linux and FreeBSD (under linux emulation so I can use the flash plugin) and it barely ever crashes -even when subjected to the massive javascript porn popup stress test.
Not actually true, BSDi pays Jordan K. Hubbard's salary - he's the release coordinator and public relations guy, as well as being one of the founding members of the project. Jordan used to receive his paycheck from Walnut Creek before the merger.
There are also a few people who work in the FreeBSD labs, I assume they were also Walnut Creek employees. No idea what their status is now.
This seems like an interesting story and all, but the editors aren't putting similar, good topics like the
non-inclusion of SGI's Apache speed patches (due to the size of the patches and patent problems) on the front page. That one definitely could use a Slashdot mention to get the community together and work on overcoming the obstacles to getting them committed. Faster Apache is more important to me than ASP support in Unix.
The old ICQ98a client isn't bad either. I'm still running that (on the odd occasion that I boot to Win98) because it supports message, chat and file transfer. And it's even less bloated than licq which uses ~8 megs of ram on my system.
The other strange coincidence is that she talks about working at a construction company, just like Ron Livingston ends up doing at the end of the movie.
Slashdot Mirror
I for one hope that they don't come to an agreement. That way we'll see Mozilla on a lot of desktops. Even if they were to include an crappy release like Netscape 6, they're AOL users, they wouldn't be able to tell the difference. Right?
Seriously though, it would be nice to see Mozilla get some marketspace this way and it is getting good enough (especially speed-wise on Windows, Linux needs tuning) that most people would be happy with it. XP is not going to be running on your P166, so the requirements are going to be fine and I doubt most AOL users upgrade all that often. From what I've heard from some Mozilla developers, the AOL folks hate Microsoft with passion, so don't be suprised if they stick with their horse rather than ally with the evil empire.
Yes, I think I confused what I thought the parent was saying with Microsoft's stance. My bad, my post was a bit OT.
I agree with Ballmer saying that government funded work should be public domain.
I still think my comments are quite relevant to the issue though. Microsoft has been claiming that the GPL destroys IP, I was trying to refute that (and perhaps it belongs in a different thread).
You're exactly right. Microsoft seems to believe that the Linux community has some obligation to make the source code available to them for easy integration.
This is simply irrational. Linux is a competitor to Microsoft, as they have stated countless times. Does Microsoft complain that IBM doesn't make the source of (current) versions of OS/2 avaible or Apple the proprietary bits. No, because they are twisting the facts.
The GPL acts as a sort of protective measure, keeping undesired competition out. If you want to use GPL code, you have to open your code base. This specifically protects intellectual property because Microsoft (and others) can't grab it wholesale and use it in their products without making their source available. Doing so would put them on even footing with the open source competition, in that the open source/free software community would also have access to their I.P. and be able to incorporate Microsoft code.
Imagine if SGI had made XFS available under a different license, ala the BSD or X11 license. Sun, HP, and IBM would be able to port the source code to their version of unix and as a result, have a better offering. But the GPL adds an additional barrier to entry, they would have to do something they have no desire to do, make their code base entirely open and free. So, they will not use the XFS code and SGI's I.P. is being used in the way they want it - that is, they will ahve access to the source of any product using XFS. That is precisely what they want, because they want to be the guys selling you hardware that uses XFS.
They are in business here and open source makes sense for business reasons, not some altruist desire.
The Progeny team is participating in the Open Packages effort. We may end up seeing ports in Linux as well.
DJB absolutely does not want any changes made to the directory structure. That is the principal reason he gives for being so anal about distribution. The packagers WANT to change the directory layout - it provides additional support for them. Having a package install in exactly the same place everywhere makes it easy to admin everywhere.
/var/qmail directory is great. Why not make the license something like the BSD license with the explicit condition that you never, ever change the directory structure and if you make changes, change the name. That way it can be the default MTA.
:) There are plenty of other academics who are a bit less unusual. He's put out lots of great software, this I will not dispute. It's rather interesting to see that he's able to write what is generally faster, more secure software as one person working alone than teams of developers.
I wholeheartly agree with this. Having everything in the
The rationale is that there are frequency upgrades to documentation, and the website will always be up to date.
That's fair but ideally your documentation should reflect the current version. Your interfaces aren't going to change after they're written and the tools will still accept the same switches as when they're released. The documentation should be functionally complete for the version that's being released.
OK. Bernstein is a little quirky. It often comes with being an academic. However, he has done a lot for free software even in just considering his cryptography court battle. And he has written a lot of software as free as QT 1.0 if not free-er. http://cr.yp.to/software.html
I think it comes with being DJB
In a perfect world, it would be nice if his license loosened up a little. But he who writes the software gets to choose the license, and DJB makes things as free as his own sense of what is right in software allows. If he were to GPL his programs, then all the distributors would alter directory structures. And he thinks that is bad. So he allows unrestricted binary distribution with the exception that the distribution has to occur EXACTLY as the tarball build would make it.
Not quite that simple, from http://cr.yp.to/qmail/dist.html:
(3) the package's creator warrants that he has made a good-faith attempt to ensure that the package behaves correctly. It is not acceptable to have qmail working differently on different machines; any variation is a bug. If there's something about a system (compiler, libraries, kernel, hardware, whatever) that changes qmail's behavior, then that platform is not supported, and you are not permitted to distribute binaries.
To me, all patches change behavior. In order to include qmail in the src tree of the bsds, you will have to make modifications to the makefiles, etc. However, he does specify that he may make exceptions if you email him, I may just have to ask him about it.
I'm not suggesting he GPL his software. Just make it a bit easier to include as the default MTA in linux and bsd (e.g. allowed _labelled_ changes). Though, I can see some people rejecting it solely on the basis of the directory layout...
Anyway, I'm beginning to see that I'm getting a bit anal about this. Just liking free software, I'd like his stuff be a bit more free (and send sendmail to the retirement home).
There should be a nice performance gain when they move to gcc 2.95.3 and start using -O2 to optimize as well.
.9 release.
See http://bugzilla.mozilla.org/show_bug.cgi?id=53486 for details.
The only reason this hasn't happened yet is they didn't want to introduce potential compiler and optimization issues right before the
Offtopic, but for anybody having problems with junkbuster breaking certain pages with mozilla, you can fix this by disabling keep-alive in the Debug -> Networking section
djb writes great code, but his license is nuts. I love qmail and have thought about it a bit and have come up with the following legitimate problems with not allowing modified redistribution.
1. You will never see qmail included as the default MTA in any free OS. Whoever is doing the packaging/migration into the distribution is going to want to make patches at some point or perhaps change the directory layout. Not to mention that it interferes with certain goals of some projects to provide completely free as in speech software.
I can understand his qualms about insecure versions being produced. Why not address this in the same way as the artistic license by making it so that only djb can produce binaries called qmail and other people would have to change the name to indicate that they are distributing modified versions (call them OpenBSD-qmail or something).
2. I can't use his libraries for my own projects. One point that he raises on his web page is that he's got a much more efficient dns resolver than the one that ships in BIND. Great, he gets to use that in qmail and his programs, but if I want to use it, I have to ensure the user has a copy of djbdns installed just so that I can make use of a small part of the program. Is linking against his libraries even allowed under his license?
The other thing that really peaves me is that he ships what I consider incomplete versions. djbdns does not include installation instructions or man pages in the tarball (the latest from his webpage). Instead, he expects you to read them off his website. This is damn inconvenient and it suggests to me that he wants to measure the usage of djbdns or some such. If so, it's a pathetic means of doing it. I hate control of information, especially in a case like this. You can man-ify the html, but why make people do extra work.
I've also heard that he doesn't comment his code. Given that he doesn't mind people writing patches, why not make it a bit easier to understand what's going on.
qmail is great and from what little experience I had with it djbdns isn't bad either. I just wish he would lighten up a bit. At least some outside patches are finding there way into djbdns but qmail could use a minor update to clarify the documentation (drop support for inetd) and include 1 or 2 of the performance patches out there.
As somebody on slashdot previously wrote, "nobody codes more paranoid than djb." Hopefully qmail 2.0 will include some cool stuff we he has it finished.
The things I've read said that currently the Mac port is the worst of the three main ones in terms of stability/performance (not a dis on Macs, but just the way Mozilla is). However, I'm suprised to hear its that bad.
BTW, the interface is pretty much fixed. You can add and remove some of the buttons, but if you want real change get new themes. New themes are available thru the view, apply themes, get new themes menu (which takes you to x.themes.org's theme site). The Lopburi flat theme is great, if the guy ever gets around to removing the text labels from the buttons.
BSDi is considerably different from FreeBSD. They both started around the same time if you count the roots of FreeBSD in 386BSD but have been developed separately.
While BSDi has borrowed some code from FreeBSD, their trees are divergent enough that when FreeBSD began work on SMPng, they had to code from the ground up (sharing some common primitives) because the BSDi code wouldn't port over to the FreeBSD kernel.
Although I am not particularly familiar with BSDi, I can assure you that they have their own developers to implement features on their own when they don't find something appropriate in Free|Open|NetBSD.
Not to mention you have your facts on Sun wrong too. They did use BSD code to create the first few versions of SunOS (pre 5.x) but at the time, the code was not free. You had to pay for a license in order to get a copy. They may have even had to pay an additional fee to produce a derived version for profit.
You have some good points, but your facts are lacking. Get a clue, fool.
If you want a window manager that's truly stripped down but still has plenty of functionality, try pwm or Ion which both make good use of the keyboard and run in about a meg of ram.
pwm somewhat resembles BlackBox without the bar at the bottom or iconify, adding the ability to attach windows together into frames for each navigation. You can move thru the frames with hotkeys, keeping the desktop more organized.
Ion takes it a bit further and forces all windows to maximize to fit a fix sized box (frame). You can have multiple programs in a given frame and can navigate thru them or move from frame to frame with mouse or keystrokes. It's truly odd and a big departure from traditional CDE or Windows knockoffs.
These are where real innovation and progress in window managing is. Making something that doesn't just look pretty or makes Windows users at home, but who would have guessed, actually help you manage windows better.
I'm not surprised at all. They'll do whatever they think they can get away with that will help them make money.
They're undercutting us by paying for Verizon the ISP's losses with Verizon the telco's profits. Since the ISP and telco are actually two separate companies on paper (Verizon Communications is a parent company of Verizon Internet Services), AFAIK they aren't supposed to share money back and forth. I have little legal knowledge, so I could be totally wrong on this and if I am my whole argument is meritless.
If I am correct, I have no way of knowing whether they are doing that, but if they aren't Verizon Internet can't be making any money. Bandwidth/admins/other costs aren't cheap enough they are able to make it on $7.50 net income. There's no way they could afford to do that for more than a few months.
I'm not suggesting they should offer us some advantage, but rather play by essentially the same rules that we have to. They charge everyone $32.50 as a "line charge", fine, then expect everyone to compete on the ISP charge portion. Don't send have the telco Verizon Internet a check every month to make up for losses.
One of the primary problems is that there is a flat rate for line charges. If Congress removed the tariff that dictates the $32.50 price, then broadband might be a lot cheaper right now. (Under the terms of the tariff, the price for ISPs with over 500,000 subscribers goes down - so it is possible that Verizon Internet is doing this, it's just damn hard to figure out what's going on).
Yeah, I don't see what any real advantages Verizon has there. We outsource our news to Supernews and I haven't heard any complaints from customers. From my experience, it's not bad, pretty reliable when I've used it. However, I haven't done many binary downloads, I believe our cap is somewhere around 256kbits so it may not be what you need.
One main advantage that they have is 24 hour tech support. Right now, we don't, though we have considered it.
As far as the other services go, they're all pretty inexpensive so we do them inhouse. It's really not that hard to keep a system running if you know what you're doing. A couple FreeBSD servers running qmail and a Windows NT machine from before I got there. We have our outages when something breaks but generally things are pretty good.
One thing I like about our setup is we use NAT, which no one else seems to do. It's rather nice since we can allow people to have multiple computers connected without using lots of public ips. Of course, NAT has it's own disadvantages but we charge $5 per static ip so it's not too bad.
If you want, I'll email you our url (who knows, you may have already tried us). I not posting my email address or our info because I prefer not to find somebody at Verizon getting ticked off at my comments.
I work as a sysadmin for a small isp in SoCal and one thing that's makes it really hard for us to get by is some of the ways in which GTE/Verizon and PacBell try and squeeze out the little guys.
For example, Verizon is dictated by law that they must charge $32.50 for the DSL line (not including ISP service). That was fine when they charged a total of $49.50 for a complete package. But in order to remove competitors, they have lowered their prices to $39.99, causing their ISP division to take a loss presumably, while they pay for it with profits from the telephone division. This makes quite hard for ISPs using Verizon to supply the physical link to make it on such a small margin.
PacBell used these tactics as well and has virtually eliminated all the competition in their areas. They set their price at $39.95 and once they had most of the market, they raised it back to $49.95. This only works when they have large cash reserves or can abuse their relationship with the telco to fund price cutting with the DSL line profits
Traditionally, Verizon offered free modems to all customers who signed up for a 1 year contract with any ISP. They've since stopped that offer as well and made it so that only their customers get the modem, while ours must pay $200. Gee, which company will the customers sign up with.
This sort of crap makes it really hard to make a profit or even stay in business. We've managed to stay alive, but a lot of other ISPs in our area have gone under. From what I've heard Verizon doesn't have particularly good service, they just win through anti-competive behavior and ISPs can't afford to sue and don't have the market share to make them compromise. Definitely bad for the consumer and us as well. Hopefully, Verizon will decide they have enough of the market and raise their prices back, allowing us to make a bit more money.
For one thing, OpenGL puts most of the code in the driver. That means that the vendor has to write a good bit of the driver whereas with Direct3D you use Microsoft's implementation and hook it into your hardware.
This has the nice benefit of making new drivers pretty fast right off the bat (see ATi's troubles getting decent RagePro GL drivers), but everybody's gotten mature enough OpenGL drivers that this doesn't matter much any more. Not to mention with the consolidation of the graphics market, no one will be writing new GL drivers, just revising their existing versions.
From what I've heard, they've been having connectivity problems :)
I had the same problem with the mouse and got around it by not using moused. Instead, just use the raw device in X and set your device to /dev/psm0 (or whatever) and Protocol to Auto.
AFAIK, this was fixed by 4.2 anyway. I don't use moused as I find little need the mouse in the console. Terminals in X or straight keyboard for the servers suits me just fine.
>Meanwhile, FreeBSD suffers from old, stodgy code written by core Unix developers with >1980's-style development practices. Sorry, but the art and science of software >engineering have improved vastly, and Linux developers are *NOT* afraid of saying, >"Okay, our implementation of xyz is flawed, let's throw it away and do it again". So >Linuxleapfrogs FreeBSD and will continue to do so.
Uh, ok, let me see if I can feed the troll here. I can count some of the things that have experienced major retrofits in the recent past:
3.0 series - implementation of SMP (not so good)
port to the Alpha architecture
complete overhaul of SCSI layer to move to cam (ftp.freesoftware.com would be nowhere near as good without this)
introduction of vinum software/hardware RAID
4.0 series - major rewrite of virtual memory by Matt Dillon (not the movie star, and I believe his work was helpful to reworking linux vm as well)
IPV6 support - all new code written by the
USB support, stable a few months before linux
move to pcm audio drivers as the default drivers for sound cards, replacing the old OSS code (which still is available for compatibility)
integration of IPfilter as an alternative to ipfw
implementation of pthreads - licq and friends wouldn't even compile in the pre 3.4 era
DRI kernel modules - can be tough to get working (voodoo 3+ w/ xf 4), but it's there if you've got the time
5.0 series - SMPng in progress (should put FreeBSD back in competitive smp territory on a par with linux 2.4 hopefully)
libh project to modularize the install process (allow X and text installers) and break the system into smaller packages
openpackages - use the same build scripts for all BSDs (not really 5.0 but still a work in progress)
ongoing work on porting FreeBSD to sparc, powerpc, and arm
kqueues - which look like a good alternative to poll() and select(), though I have no experience with them
So I don't know what the heck you're talking about. Sure, Linux tends to move things into the stable branch more quickly than BSD and there are a lot more releases, but we get along just fine. From what I've heard, the FreeBSD kernel is a bit more elegant than the Linux kernel because code is less likely to be included unless it meets the committers' standards.
Linux development moves at a breakneck pace, FreeBSD is more conservative. What does that tell us? Absolutely nothing.
This is a bit of a flame, but some of the reasons that there haven't been as drastic of changes in BSD as Linux is that some systems have been high quality from the get-go. They drew on well-tested mature code (BSD Lite) that had already had a lot of the bugs worked out.
How many times has the Linux TCP/IP stack been reworked, like 4? The only place where I can see a halfway coherent argument is in the filesystem. It would be nice to see a journaling filesystem that would give better metadata performance than FFS + softupdates, but all the implementations are under the GPL. As it is FFS does fairly well and is more reboot tolerant than ext2 and until recently had Linux beat for large (> 2gig) files.
FreeBSD puts out a quality, coherent system that is stable and performs quite well on uniprocessor systems (and soon SMP). So long as they keep doing that, I'll be a devoted FreeBSD (and Debian) user.
It appears that the real problem may be with how gcc and mozilla are interacting. Please see this status update, in which it states that gcc is generating code that is twice as large as what Visual C++ is generating. If Linux builds are executing twice as many instructions, of course the Windows builds will be faster. Further follow-ups don't show much progress other than that switching from egcs to gcc 2.95.2 reduced size by about 5%.
Unfortunately, using gcc 2.95.2 requires a libc upgrade. I just installed the gcc 3.0 3/20/01 snapshot and am going to try a build tonight to see if it makes much difference.
Since pavlov was moved to help out with libpr0n, I haven't been seeing many updates about general Linux performance. Back in November thru early January, this was one of the main things the Footprint team was working on. pavlov seemed to be one of the most knowledgeable guys in the area and there were several people assigned to it. It seems they all got pulled elsewhere.
To really make Mozilla work well in unix, they need to put the focus back on speed and also take a look at the memory leaks that plague the linux version. After browsing for a few hours, Mozilla tends to swell up from around 20 megs of ram to 50 or 60. They had some graphs profiling this problem, but those seem to have stopped being updated as well.
On the whole, I find Mozilla pretty darn useable at 0.8.1. I run it on Linux and FreeBSD (under linux emulation so I can use the flash plugin) and it barely ever crashes -even when subjected to the massive javascript porn popup stress test.
Nah, it looks like one of those is their mom. Living at home doesn't count.
Not actually true, BSDi pays Jordan K. Hubbard's salary - he's the release coordinator and public relations guy, as well as being one of the founding members of the project. Jordan used to receive his paycheck from Walnut Creek before the merger.
There are also a few people who work in the FreeBSD labs, I assume they were also Walnut Creek employees. No idea what their status is now.
This seems like an interesting story and all, but the editors aren't putting similar, good topics like the non-inclusion of SGI's Apache speed patches (due to the size of the patches and patent problems) on the front page. That one definitely could use a Slashdot mention to get the community together and work on overcoming the obstacles to getting them committed. Faster Apache is more important to me than ASP support in Unix.
The old ICQ98a client isn't bad either. I'm still running that (on the odd occasion that I boot to Win98) because it supports message, chat and file transfer. And it's even less bloated than licq which uses ~8 megs of ram on my system.
There was an article on daemonnews about this and apparently NVidia has done some work to get the Detonator drivers ported to FreeBSD.
_ id=1530
http://daily.daemonnews.org/view_story.php3?story
My advice is to sign the petition and hope they listen and finish up the work.
The other strange coincidence is that she talks about working at a construction company, just like Ron Livingston ends up doing at the end of the movie.