Slashdot Mirror
glibc 2.1 is out
GTM writes "I have just seen this on the main page of LinuxHQ, and haven't even taken time to download it yet : after 2 years of development, the GNU C Library 2.1 is available. Nice to see it appear short after the release of the 2.2 Linux kernel. "
And I just finished downloading egcs-1.1.1 which should be used for compiling it. Apparently gcc-2.7 will not work.
Wow. Linux, Samba, Gnome, KDE, and now glibc. All we need now is for Xfree86 4.0 to come out in about 6 months time and we'll have a true linux upgrade here.
I'd now like to see linux 2.2.2 come out, preferably without the bug that causes the system to hang while copying files to a vfat floppy. On my computer 2.2.1 hang twice while I was trying to copy about 15 files to /mnt/floppy.
Everytime I try to update libc on my system, I end up screwing it up so badly that even ls won't work. Bleh.
Anyone know where there's some _GOOD_ documentation on this procedure?
Everytime I try to update libc on my system, I end up screwing it up so badly that even ls won't work. Bleh.
As the link in the article states, it's a terribly complicated procedure that really shouldn't be attempted without a lot of knowledge and patience.
At the risk of going against the party line... why not just wait for the distribution vendors to update their distributions? Unless you have something that's horribly broken w/ glibc 2.0, there's really no burning pressing need to upgrade to 2.1. Sure, being on the "bleeding edge" is a big ego boost, but royally screwing your system because you didn't quite get the upgrade wrong does very little for the ego.
Funny you should mention this. /usr/lib/libgtk.so.1 (0x40005000) /usr/lib/libgdk.so.1 (0x400ad000) /usr/lib/libglib.so.1 (0x400cd000) /usr/X11R6/lib/libXext.so.6 (0x400da000) /usr/X11R6/lib/libX11.so.6 (0x400e5000) /lib/libm.so.6 (0x4017c000) /lib/libc.so.6 (0x40195000) /usr/X11R6/lib/libXi.so.6 (0x4023a000) /lib/ld-linux.so.2 (0x00000000)
I've been trying to get gimp(any gimp) to work with my system.(rh5.1)
output of ldd:
libgtk.so.1 =>
libgdk.so.1 =>
libglib.so.1 =>
libXext.so.6 =>
libX11.so.6 =>
libm.so.6 =>
libc.so.6 =>
libXi.so.6 =>
/lib/ld-linux.so.2 =>
Trying to run gimp:
gimp: error in loading shared libraries
: undefined symbol: __register_frame_info
I have glibc-2.0.7.29
XFree86-lib 3.3.2.3-25
I believe the problem lies in ld-linux.so.2
Anyone else have any ideas?
Does anyone care to summarize improvements in 2.1 over 2.0? The package has more than doubled in size. Is this bloat or added functionality? How about speed and stability? Can we havesome benchmarks, please? Anyway, great to have linux-2.2 and glibc-2.1. Now we're only waiting for GNU/Hurd 0.3.
Does anyone care to summarize improvements in 2.1 over 2.0? The package has more than doubled in size. Is this bloat or added functionality? How about speed and stability? Can we havesome benchmarks, please? Anyway, great to have linux-2.2 and glibc-2.1. Now we're only waiting for GNU/Hurd 0.3. Sorry, if this got posted twice!
This will give Slackware a chance to redeem itself and move to this new glibc for the next version.
There's no doubt that Slackware overall is the best distribution IMHO (my opinion - no one else - no flames needed!) and having the latest libc would just top that off for me.
"apt-get update && apt-get upgrade" ...
forgive my ignorance, what does that mean?
``Lesser GPL'' is a plan for the next version of LGPL, which takes more thought and effort than changine one word in a file. Don't expect to see it happening so soon after it is announced, and don't make stupid comments like that. I have no clue why RMS would get flamed for wanting to rename something to better describe what it is, rather than what it has typically been used for...
- RF (dfelker@cnu.edu)
You will need this: procps-990130
Likewise.
I forwarded this info to Alan Cox, hopefully he will know what to do with it.
If you really want to do libc5->libc6 by hand,
the FAQ in the released kit has pointers to some
HOWTOs on the subject. Also, go dig up the HOWTOs
on moving from a.out to ELF; it's actually kinda
similar. You will need to recompile most everything
to get any benefit.
Moving from 2.0.x to 2.1 is fairly easy, the big
gotcha is that you need to rebuild libstdc++,
libncurses, and maybe others (hard to say) if you
want to compile new stuff. Old binaries will just work.
GLIBC FAQ, 2.8 ::
{ZW} Glibc on one of these systems was compiled with gcc 2.7 or 2.8, the other with egcs (any version). Egcs has functions in its internal
`libgcc.a' to support exception handling with C++.
They are linked into any program or dynamic library compiled with egcs, whether it needs them or not. Dynamic libraries then turn around and export those functions again unless special steps are taken to prevent them.
When you link your program, it resolves its references to the exception functions to the ones exported accidentally by libc.so. That works fine as long as libc has those functions. On the other system, libc doesn't have those functions because it was compiled by gcc 2.8, and you get undefined
symbol errors. The symbols in question are named things like `__register_frame_info'.
For glibc 2.0, the workaround is to not compile libc with egcs. We've also
incorporated a patch which should prevent the EH functions sneaking into
libc. It doesn't matter what compiler you use to compile your program.
For glibc 2.1, we've chosen to do it the other way around: libc.so explicitly provides the EH functions. This is to prevent other shared
libraries from doing it. You must therefore compile glibc 2.1 with EGCS.
Again, it doesn't matter what compiler you use for your programs.
2.1 up and and even runn... *oops, oh sh***
Most of the size increase is localization data,
which used to be distributed separately. The
announcement has a list of improvements, but the
big ones are support for kernel 2.2 features,
ELF symbol versioning (means we never have to bump
libc's major version number again *knock wood*),
thread-safe dynamic linker, partial support for
C9x, and piles and piles of bugfixes. We think
it's faster, but I don't have hard numbers.
Will we now have different RPMS for, Red Hat 4.x, RedHat 5.0, RedHat 5.1, RedHat 6.0, SuSE 5.0, SuSE 6.0???
Lesser GPL is the term that Richard Stallman recently came up with to replace Library GPL. Check out the archive.
I don't think you're right about what can and can not be linked to GPL'd libraries. I think only GPL'd apps are allowed to link to GPL'd libraries. In the GPL there's only a distinction made between GPL'd and not-GPL'd software, not between free software and non-free software.
And yes, RMS wants libc to be LGPL'd because there already are commercial implementations of libc.
but the question was about the difference between glibc-2.1 and glibc-2.0 :)
Then why doesn't he rename the "GPL license" to be the "viral GPL license" to better describe it?
I have gcc-2.7.2 and egcs 1.0.8 The problems I've been having are with any version(fairly recent).
Binary or compiled(gimp). If I'm understanding the ldd output. the 0x00000 means that's the shared library it's having problems with. I guess I'll have to get the source and compile that library.
After that I'll then re-tackle getting lyx to work.
This upgrade breaks apt. I don't know what causes the problem, but JGG has been notified. I would advise holding off until the problem is resolved. Downgrading these packages was a pain in the butt.
Why dont you stop and learn to read. Then go back and make another pass at RMS's article. He said that it was a stragitic decision to be made by the developer. He thinks that LGPL is good, but he think's it being a bit overused. If you choose wisely you can help free software alot more.
I'm running potato. Others have seen the problem in slink, I think. This is what happened when trying to run apt after the upgrade:
/var/state/apt/lists/llug.sep.bnl.gov_debian_dists _unstable_main_binary-i386_Packages is too long.(2)
root@oppy:/usr/doc/libc6# apt-get update
Get http://llug.sep.bnl.gov unstable/contrib Packages
Get http://llug.sep.bnl.gov unstable/main Packages
Get http://llug.sep.bnl.gov unstable/non-free Packages
Fetched 644k in 3m7s (3430b/s)
Updating package file cache...
E: Line 3 in package file
I'm using an up-to-date potato system with apt-0.1.10.
Apparently you've never had
It's a good idea, but it's been implemented very wrong.
No, it is a bug in glibc2.1, aptv3 does not use the C++ IO streams for parsing package files, aptv1 does. They seem to break somehow.
and I have an error when trying to compile mimelib for kde 1.1 that looks somewhat similar:
.libs/libmimelib.so.1.0.0: undefined versioned symbol name __register_frame_info@@GLIBC_2.0
/usr/bin/ld:
/usr/bin/ld: failed to set dynamic section sizes: Bad value
What is a *versioned* symbol name, and how do I fix this?
jce2@po.cwru.edu
Every change to Linux results in more and more problems. Its why real operating systems change very infrequently. OS's like Solaris, AIX and NT. Thier users sleep soundly at night without the nightmare fear of the hacker community "fixing" a non-problem or bloating out the libraries again. Just like now.
The new release of glibc is just "jobs for the boys" GPL developers changing things just to keep themselves employed.
1999 will be the year when IBM, Dell, Compaq and others expose this schit and we can get back to stability.
Keep those changes coming people. Its the only way you can sell your "support" contract. Ya know, break things on purpose and charge for the fixes. Just like M$
Look, just because Linux development is lightning fast doesn't mean you HAVE to update. Slackware is still using libc 5 and it still rocks, poeple complain about it, but most of them could hardly tell the difference. This is a development thing, you don't HAVE to use Linux 2.2, you don't HAVE to use KDE 1.1. You upgrade every piece of software that gets a newer version on your Windows system? Wow! Notepad 1.3-BETA is out! I MUST have it! Or do you like bug fixes to come out after 3 years (ie. Windows 95 -> 98).
Well, I hope you get enlighted real soon..
signed: the anonymous wimp
Things like Unix98 pty support, realtime signals, async I/O, and the new system calls. These features are in kernel 2.2, but they don't get used unless libc knows about them. For example, system call #183 is getcwd(), but an old libc won't use it when you call getcwd() in your program.
The kernel is completely self-contained, so it doesn't matter which C library was in use when you compiled it.
No, you can't use gcc 2.8. There are binary incompatibilities between its implementation of C++ exception handling and egcs's. If you use egcs to compile libc, all C++ binaries (should) work. If you use gcc 2.8, many C++ binaries will fail to work. This is the `__register_frame_info problem' that's being discussed a few threads down.
You can use either egcs 1.0 or egcs 1.1, and it doesn't matter what compiler you use to compile other stuff.
Note that this is different from glibc 2.0, where you need to use gcc 2.7 to avoid the problem.
And once a security hole is found in your proprietary operating system, its around for months, and on EVERY SYSTEM the OS is installed? Seems like hacking on any proprietary operating systems would be like mining a veritable gold mine.
Yeah, I like your logic
Hehehee thats preaty funny.
...but maybe it's not. So, free clue:
You don't *have* to upgrade. I'm still running
RH5.0 (with patches) on my laptop and everything's
just fine. Running RH5.1 (with patches) on a
Web/FTP server and everything's just fine.
You're obviously far too used to the Microsoft
mindset that usually affects only clueless end-
users and PHBs: that when a latest & greatest
comes out, everybody who's anybody has to upgrade
*right* *away* or technology will leave them
behind!
Solaris and NT in the same sentence? Maybe it's
a troll after all.
Name one thing you've done for the Linux/Open Source/etc movement, then match it against what ESR has done, then repost. On second thought, don't re-post. Save slashdot space for someone who can post intelligently.
Guys like YOU are why this movement is losing momentum.
If you want something stable, get the stable releases of some distribution like Debian or RedHat. They are (supposed to be) stable (and badly lagged).
But if you want something that works, speed, and fun, get the unstable releases! What's a hacker's life without a bit of fun?
> Real operating systems change very infrequently, ... like NT.
>
NT is doubling the size of its source code from version 4 to version 5.
> 1999 will be the year when IBM, Dell, Compaq and
> others expose this schit and we can get back to
> stability.
Thus far 1999 has been the year when IBM, Dell, Compaq and others exposed NT schit and went to Linux stability.
> Ya know, break things on purpose and charge for
> the fixes.
Glibc is distributed for free. No fixes are ever charged for. As opposed to the "real" operating systems your brought up.
Do you even bother to think about this shit before posting it? You remind of the Dilbert boss. I can imagine you muttering to yourself "It almost makes me wonder what a 'glibc' is..." You apparently don't know, because you think glibc's an operating system and AIX, NT are better...
Moron, moron...
Eep! I've offended with your statements, but because I've lost my password and changed my E Mail I can't fairly reply. Yes, I support getting rid of ACs, they're more often than not trolls, but I think this post is not an example why. This post has SOME intelligence to it, and maybe a few worthwhile points, it's not "linux sucks balls, NT forever" like most AC posts. Hopefully we can enlighten this person, who probably is not aiming to start a flame war (or that's only a secondary goal) and more trying to point a flaw in the current OSS system.
I don't understand why he didn't spend five minutes getting an account though.
schmoko (no password, changed email)
I'm only speaking for myself here, but who uses floppies anymore? Apart from the light blinking on startup, I don't know if my floppy drive even works yet. (no disks, no use, it's just there)
I suppose they serve someone's use, though..
If it doesn't invade on privacy, then how does it help? The whole idea behind it is that people will be forced to put their name behind what they write; but if they're still anonymous, then it doesn't help any. If you really want all the trolls to set up accounts with fuck.you@bill.com email addresses and sigs 10 lines long, fine with me. Personally I find the idea of accounts a waste of resources...we can (a) be anonymous with fake information stored on a data base, or (b) be anonymous.
muhahaha
eat me.
Show the poster's Fully-Qualified-Domain-Name (i.e., xyz123.bleh.domain.net) and possibally (sp?) even their IP address.
I still think AC is fine, Rob just needs to update the code to show the posters FQDN.
Gotta agree. ESR really just has a big mouth. And there's no reason that everyone has to blindly bow down before him every time he shits a brick on commercial software or some Microsoft documents. I'm not a Microsoft supportee, I'm 100% Linux all the way. But I do think that there should be some software in the world that costs money. Otherwise, the free software never has anything to equate to.
Just my opinion on the whole ESR+OpenSource+RMS "thing".
Well i -had- a standard redhat 5.2 system, now ive upgraded to linux-2.2.1, glibc-2.1 and to top it off, a full egcs set (overwriten the old gcc). And my first conclusions... xwindows starts about 30% faster, netscape finaly runs on decent speeds! i mean seriously better, untarring, listing dir's etc, it flies!!! not that i had crapy hardware before, its a p II 400, with 256 megs ram and a u2w scsi disk, but for the kind of hardware i have, it was reacting just a little to slow, no more!!! this is a definate advise to do the same!
:) After a few minutes of compiling (read: hours for the 3) you'll be flying!
Hints to install:
install the kernel first (2.2.1), then egcs-1.1.1 (from egcs.cygnus.com) and then the glibc-2.1 it will recognise the compiler+kernel, and go for it, it prompts for a small change or 2, make it so, and before all read the INSTALL's
Congrats on a great release guys!
Being the brave soul that I am, I chose to install it over my previous glibc. And it worked! Everything looked great. And then the complaints rolled in. No telnet, no ftp, no finger, no biff. Everything gives me this error: /lib /usr/lib and /usr/include, but I'm going to have to boot a microdistribution (meaning physical access, which isn't easy) to restore from my backups. Why can't this be easier?
getpeername: Socket operation on non-socket
I can't find the source for in.telnetd anywhere to recompile it (debian didn't have it, ?) And you are incredibly screwed if you want to remove it. There's a long story here that I'm omitting. I backed up my
Look at it another way,
...development is happening all the time, Windowz is a controlled by one company...who says when major updates are done, Development of the main system is slower. Look at the problems coming out of Windows2000 compatibility with NT products? Not everything is Perfect esp MS-bi products,.
Linux is Open Source
Essentially time-delayed bug fixing....OSR1-2-3 etc... we have patches available in hours (most of the time) if not.. the a few days after a release.
Sure glibc2.1 has a few problems with some libraries, but this is an advancement..evolving! It is showing our development continues 24hrs/7days-week.
we the OSS community evolve, our source code matures...no so for windows.
I hope you like the win9x kernel...cause its going to be around for a few years yet.
The C library should be stablized and then never changed unless there is a pretty damn good reason. Changing it only causes headaches for anybody trying to put out software on these platforms, it's like trying to hit a moving target.
A bit off topic:
Since 2.2 my network is killed (red light at the hub) when I start Linux, I simply can't use it.
Digital WS (PII) with Tulip card driver.
2.1.131 runs fine (but sometimes the kernel panicks when mounting the SCSI ZIP)
C is a moving target. Get over it.
It's a lot better than having mature (read: old and buggy) libraries, like Win32. Wow, how many apps brake when going from Win3.11-win32s to Win95-win32 to WinNT-win32? I'd rather have a library that worked as documented, instead of working around documented bugs.
Well, much to my surprise, invoking these services manually even on a healthy system will generate this error. But for some reason, under my same configuration, when the services (through inetd through tcpd) are used, they immediately disconnect.
Now I am really confused. Should we use egcs or not when going from glibc-2.0 to 2.1? /lib/libgdbm.so.1: undefined symbol: _xstat.
I compiled with egcs and installed now now have many woes with programs failing to start with ld errors: eg
/usr/local/apache/sbin/httpd: error in loading shared libraries:
I am having touble finding a version of gdbm to that compiles shared.
Just add the following line to your /etc/apt/sources.list
deb http://www.debian.org/~espy/ glibc-2.1/binary-i386/
then run apt-get update; apt-get dist-upgrade
The packages will be signed when they are uploaded into the main archive. espy happens to be the libc maintainer for debian, If you can't trust these packages you might as well not trust anything you download off the net.
Yeah, but you may notice a small disparity in those patchlevel numbers... It isn't 2.0.1 you're running there, after all. 2.2.* will get there... it just needs a little time to shake the bugs out.
Yeah... the machine that I've been using for kernels from 2.1.x up through 2.2.1 hasn't even got a floppy drive, or any other removable media, for that matter, and never has. I booted the thing the first time off a 120M drive I pulled out of another Linux box that I'd upgraded to a 2.1G, changed hostnames and IPs and the like, and off she went... She communicates with the rest of the world over ARCnet... if for some reason I need some of her data on a floppy, I just transfer it to another machine and to a floppy from there.
Posted by irritron:
Funny I am using gcc version egcs-2.90.29 980515 (egcs-1.0.3 release) and both Window Maker and the
Gimp built, and run fine. But i have seen some problems with egcs and C++ over here. OffiX would
not compile for me.
Posted by Atma David Stormfighter:
hey... you managed to upgrade from libc5->glibc2? I need help with it, I've tried the walk-throughs and they always die on compiling the compiler to use glibc2 instead of libc5. I'm using a nearly unmodified Slackware 3.5 distribution, and I seriously need glibc2, since GTK+-1.1.x really wants it, and ninety percent of the stuff I wanna use needs GTK+. q= So... if you can help me, please send it my way, removing the NO-SPAM. from my addy, of course. q=
Posted by Myrdraal:
/glibc/Glibc2-HOWTO.html - However, it is a little bit outdated. You no longer need the localedata addon with glibc 2.1, so you can just ignore the references to that.
A good place for glibc compiling information is the glibc HOWTO. http://www.imaxx.net/~thrytis
-Myrdraal
So does anyone know if this means that PPC users can start R4-->R5 upgrade? This is what we've all been waiting for, right?
Gee, just can't wait to reformat/reinstall....
Guess RMS got flamed pretty hard after that Lesser GPL quip.
Oh yea, we were talking about the GPL, not the LGPL there. Sorry.
" GNU GENERAL PUBLIC LICENSE
Version 2, June 1991
...
...
...
10. If you wish to incorporate parts of the Program into other free
programs whose distribution conditions are different, write to the author
to ask for permission....
[the program program would there be the GPL'ed programm... if the author of the program with the "other" dist conditions would have to ask the author for permisson to inorporate part of the GPL programm is not clear.]
...
...
...
[At the end of the license:]
This General Public License does not permit incorporating your program into
proprietary programs. If your program is a subroutine library, you may
consider it more useful to permit linking proprietary applications with the
library. If this is what you want to do, use the GNU Library General
Public License instead of this License."
That is you cant link it to proprietary programs. That would allow programs under other licenses wich count as "non propriteary" to link to it. Im not sure tho, I only skimmed thru the GPL this time.
Yes, that was what I was saying too, perhaps it was not clear. That is, you should be able to link some code to GPL'ed code, as long as its free. Not gratis, really really free. I also pointed out earlier that there were not many such licenses other than the GPL.
yabber, yabber, yabber. the anonymous pests of slashdot descend on yet another post that was veering dangerously close to the Other Side of the Coin. since you have no intention of discussing the matter without resorting to name-calling and the arrogant belief that your comments are more worthy of attention than someone else's, be quiet and let the rational people talk in peace . . .
I downloaded and installed a pre-2.1 release and everything worked fine except StarOffice (what a surprise - the only application I don't have the source code for is the one that breaks). This is fairly critical for me as all my accounts are in StarOffice, and I wouldn't want to lose it just yet.
--
Matt. Want XML + Apache + Stylesheets? Get AxKit.
Really - this must affect a huge amount of people - and RH6 will be out soon enough, and SO won't run on it. I hope something can be done about this. I'd rather not lose all my accounts .
--
Matt. Want XML + Apache + Stylesheets? Get AxKit.
Hmmm. This should be interesting. I fully expect some networking code (esp. multicast routers) to need patching for this. Just thought - compiling glibc-2.1's going to be painfully slow, on my machine. Does anyone have a pre-compiled set (pref. using a very recent PGCC)?
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
Let's see... I'll start by saying that there are a
few THOUSAND technical websites that Linux can
reach and NT cannot, and that glibc 2.1 will help.
Some strange, mysterious, proprietary protocol? No.
Just IPv6, which is still pre-alpha for Windows.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
"apt-get update && apt-get upgrade'
Hurry up and jump on the individualist bandwagon!
Yes. The thing is, when you upgrade libc, you have to recompile everything that depends on it. It's much better to use a package system where each package maintainer is responsible for recompiling his/her own package and everyone can download the updates. I know that I can't really leave my computer compiling everything from source..Gnome takes all morning, the whole system would probably take a week or so and be broken for a month. :-)
Daniel
Hurry up and jump on the individualist bandwagon!
I wasn't around then but I heard somewhere it was the same with 2.0.[early] ?
If you like SysVinit and PAM, choose RedHat.
I, personally hate both of them. And I love Slackware for being the last remaining distro
with the old Linux spirit : take the best from
BSD and SysV.
-- "Life is easier since I have excluded JonKatz stories from my homepage"
> Every change to Linux results in more and more problems. Its why real operating systems change
Are you sane? Every change to linux adds features and fixes bugs. And by the way, this is glibc (the gnu c library) not linux (the kernel put out by wonderful such as Linus and Alan Cox). Please don't be ignorant.
> very infrequently. OS's like Solaris, AIX and NT. Thier users sleep soundly at night without
> the nightmare fear of the hacker community "fixing" a non-problem or bloating out the
> libraries again. Just like now.
In this sentance, you have included NT with the likes of Solaris and AIX and called them all "Real" operating systems. I can understand Solaris and AIX being "Real" OSes, but NT? What criteria have you used to decide this? It certainly couldn't be based on either speed, stability, or security. If you had bothered to go to the site listed in the posting, you would have seen the large list of bugfixes and useful added features. And whether NT users sleep soundly at night or not, it pribably has nothing to do with their choice of OS.
> The new release of glibc is just "jobs for the boys" GPL developers changing things just to
> keep themselves employed.
This is laughable. "jobs for the boys"? Most of the people who worked on this release, with the exception of some at Cygnus I believe, did this work for _free_ outside their normal job. To claim other wise is downright stupid.
> 1999 will be the year when IBM, Dell, Compaq and others expose this schit and we can get back to stability.
Oh, this explains all of their recent commitments to sell linux. Unless by "back to stability" you mean away from NT. In that context the sentance makes sense.
> Keep those changes coming people. Its the only way you can sell your "support" contract. Ya
> know, break things on purpose and charge for the fixes. Just like M$
No, not like M$. M$ doesn't fix anything, but charges for it anyways.
Rob, _please_ get rid of ACs. I was of two minds about this until very recently, but I've come to the conclusion that most of the time they just are stupid. And _please_ no one give me that crap about privacy. Thats just bull$!@#. I have an account, and it didn't hurt to get it.
I have heard that the latest ANSI C specification specifys some string handling functions. On the page referenced by the link on the story above, it mentions that new string functions added. Does anybody know if these strings that have been added are the new ANSI C Strings? Or am I way off track here?
I have to admit, I did speak rashly. The FQDN idea is a good one.
"Every change to Linux results in more and more problems. Its why real operating systems change very infrequently."
An articulate observation! That's why everyone should use MS-DOS 3.2. Without those annoying upgrades like multitasking, security fixes, and support for cdroms or hard drives bigger than 500 meg. It's so simple what could go wrong?
Aside from your grasping for straws (NT unchanged?), do you do have a good point. Support isn't going to be what drives the money in the relm of open source and GPLed software. It will be hardware manufacturers advertising a better option and writing drivers so the platform independant APIs/kernel can make their superior hardware run circles around the the status quo.
"1999 will be the year when IBM, Dell, Compaq and others expose this schit and we can get back to stability."
Who knows about Dell, they will likely play both sides off the middle until a winner becomes clear.
IBM wants to see M$ gone so they can take over and are our friends(at least in the short term), and Compaq is one of those companies with the superior hardware (Alpha) who can use the intrapolarity of Linux as a long term viability argument.
That ends our lesson on how to give both sides of an argument so you don't sound like a crazy fanatic. Incidentally do you stand to lose a job or stock value if M$ market share goes down?
Novel theory: Modern Man evolved from psychopath
What (if any) changes would I see from a user's perspective, using a system built around this library as versus one built around the earler version?
StarOffice is broken. It uses internal libc symbols which aren't supposed to be used by applications, because they are constantly subject to change. The release note says that mechanisms have been put in place to avoid this in the future. However, this cannot fix older, broken applications. Thus, StarOffice will not work with glibc 2.1. Yet another example of why source code is important.
--
This sounds like some M$ minion trolling as hard as they can.
The big difference between so called commerical "real" OS and the open "free" OS, is choice and certainary.
Choice as and when you wish upgrade to the next stable version you can. (Personally I'm waiting for 2.2.x, x > 4, before I upgrade the kernal.)
Even when you don't chose to upgrade at every announcement, becuase of the very active and public development you can feel that things are always improving.
This sense of movement is typified by the 'ld core' bug in the early 2.2.x kernal. Discovered and fixed within half a day. I can image a "real" OS vendor, spending the first month before getting the fix through both marketing and the layers.
This all provides a deep certainary, and security, that the underlying OS is the best it can be and, if not, will be improved quickly and in a public manner.
es
I understand that is a microkernel design, but other than that what is the point? Linux is already here and although it may not be perfect I can't think of a better system for the hardware it runs on. From what I understand there are a few things that FreeBSD does better or more efficiently, but overall linux is still a better system. So now we have hurd. I'm not even going to go as far as to make fun of it with the letter T. Whoever came up with the name clearly didn't think about how easily it could be maligned. I've never used Hurd, I've only read about it. It may have the foundations of a great system, but is it necessary? I'm not a kernel hacker so I don't know how difficult it would be to evolve the linux kernel to the point that it had the features that hurd promises. I just think that the effort to develop hurd would be better spend developing and improving linux. We don't have an infinite supply of programming talent at our disposal.
Don't be affraid! 2.2.0 (yeah, I know..need to upgrade..) is VERY stable, and 3-4X faster if you're running SMP... there are little buggies here and there, but no show stoppers.. I would suspect that the majority of the bugs are in little used/adbused sections of the kernel. The sooner everyone jumps in there and tries is, the sooner all of the bugs get found! Just thik about it this way, all you have to do to roll back is change a symbolic link and reboot from a different kernel...
RMS basically wants all developers to use the GPL for libraries, not the LGPL. The difference being that you can't even link your app against a GPL library without GPL'ing your own app. The rationale behind this is that he believes this will give free software an edge over proprietary software, which will be denied the use of superior GPL'd libs.
It's a matter of perspective. RMS sees proprietary software using LGPL libraries as a free ride that isn't returning anything to free software. I'm more inclined to agree with ESR though and say that use of LGPL code in proprietary software gives free software leverage over the commercial code by requiring it to target a specific API. Thus allowing the free software developers to set the standard. It's rather like the power Microsoft wields, with the crucial difference: it's all open and free.
I've finally had it: until slashdot gets article moderation, I am not coming back.
I transfer files betweeen home and lab by the floppy/IP protocol :) or, more accurately, floppy/tar. who use filesystems on floppies anymore?
Yep, glibc 2.1 is supposed to be able to run all apps compiled against 2.0, unmodified. And it's a *major* upgrade, almost comparable to the kernel jumping from 2.0.x to 2.2.x. it's true that we're getting new major versions of a lot of things all in a row... and in a matter of months we get the GNOME release, XFree 4, and Mozilla!
ARGH! did the guys at StarDiv *have* to break it so bad? more confirmation that the process of open development does produce better code...
"...Why don't you try something out yourself before questioning its value, and contribute kernel patches before using talking about allocating kernel hacker programming talent. You'll have more of a leg to stand on and you just might learn something along the way..."
Don't be so upset...maybe this guy is a manager. In which case, his actions of making uninformed decisions are actually par for the course.
"...Beer..."
: undefined symbol: __register_frame_info
I was missing the same symbols when installing KDE on a rh5.0 system. You need libstdc++-2.8.0. The rpm from rh5.2 should drop right in.
simply set your threshold to 1 and you wont see the anonymous posts. warning: untested advice.
What, no ZIP Disk?
-- Keith
This sig is the express property of someone.
I just installed glibc2.1 on debian slink (debian version 2.1). and it breaks most binaries from what i have seen do to the following library.
libwcsmbs.so.0.0 is broken by the glibc2.1 install and will force you to having to reinstall slink.
I think what you'll see by going for the GPL for libraries too, is that you'll have multiple implementations of most interesting libraries - one GPL'd, and at least one under another license.
So what advantage would the free software world get? A lot of duplicated work, and less focus on filling new niches, while wealthy corporations easily can afford to buy third competing third party libraries to link their code against?
I don't want lots of apps linked against different libraries with the same functionality just because they're under different licenses.