[clc-devel] CRUX-2.2-devtest2
Hi, I uploaded a new developer test release at http://crux.nu/~sip/devtest2 Regards, Simone -- Simone Rota WEB : http://www.varlock.com Bergamo, Italy MAIL: sip@varlock.com
I limited my $COLLECTIONS to core, because x11 doesn't seem to want to build on my CRUX 2.1 system (a glibc too old? Maybe it doesn't build with gcc3?). Here is the output for a bootstrap of core: devtest2 # make bootstrap Bootstrap started (2006-02-17 14:43:49) - Stage 0 (2006-02-17 14:43:49) make[1]: Entering directory `/home/media/software/CRUX_Buildkit/devtest2' =======> ERROR: Md5sum mismatch found: =======> ERROR: Building '/usr/pkgs/db#4.4.20-1.pkg.tar.gz' failed. =======> ERROR: Md5sum mismatch found: =======> ERROR: Building '/usr/pkgs/flex#2.5.31-2.pkg.tar.gz' failed. =======> ERROR: Md5sum mismatch found: =======> ERROR: Building '/usr/pkgs/grub#0.95-3.pkg.tar.gz' failed. =======> ERROR: Md5sum mismatch found: =======> ERROR: Building '/usr/pkgs/pciutils#2.2.1-2.pkg.tar.gz' failed. =======> ERROR: Md5sum mismatch found: =======> ERROR: Building '/usr/pkgs/slocate#2.7-3.pkg.tar.gz' failed. CFLAGS="-O2 -march=i686 -pipe", btw, on a Pentium 4 with good RAM. I wonder if the sources are just down, or what? I wonder if something like this: http://www.coralcdn.org/ could prevent such incidents? Cheers, Nick
On Fri, Feb 17, 2006 at 04:19:12AM +0100, Simone Rota wrote:
Hi,
I uploaded a new developer test release at http://crux.nu/~sip/devtest2
The awk part of the man/makewhatis script has a silly bug. Fixed in [1] and sent to upstream too. Greetings Juergen [1] http://jue.li/crux/files/man-1.6c-1.prt.tar.gz -- Juergen Daubert | mailto:jue@jue.li Korb, Germany | http://jue.li/crux
On Fri, Feb 17, 2006 at 04:19:12AM +0100, Simone Rota wrote:
Hi,
Hello,
I uploaded a new developer test release at http://crux.nu/~sip/devtest2
To be honest, I'm not completely happy with the current divison of core/opt on the cd. I'd suggest the following changes: 1. Move core -> opt cdrtools cvs gdb linux-identd pcmcia-cs rp-pppoe wireless-tools wvdial 2. Add to opt fetchmail procmail mc kind regards Juergen -- Juergen Daubert | mailto:jue@jue.li Korb, Germany | http://jue.li/crux
On 02/18/06 15:29 Juergen Daubert wrote:
To be honest, I'm not completely happy with the current divison of core/opt on the cd. I'd suggest the following changes:
1. Move core -> opt cdrtools cvs gdb linux-identd pcmcia-cs rp-pppoe wireless-tools wvdial
2. Add to opt fetchmail procmail mc
Sounds ok to me, maybe with the only exception of mc that I don't use so often, but it won't hurt me too much. Since we're on topic, any other suggestion regarding the port to include/remove? In particular: - Are we ok with the core/opt separation? - Additional opt ports to ship on the iso? - Do we need a contrib collection on the iso? (even if not needed, it could speed up remastering of custom images). If anybody has some suggestion please comment, so we can avoid the risk of shipping 2.2 without the BASICS ;) Regards, Simone -- Simone Rota WEB : http://www.varlock.com Bergamo, Italy MAIL: sip@varlock.com
Simone Rota wrote:
Since we're on topic, any other suggestion regarding the port to include/remove? In particular:
- Are we ok with the core/opt separation? - Additional opt ports to ship on the iso? - Do we need a contrib collection on the iso? (even if not needed, it could speed up remastering of custom images).
If anybody has some suggestion please comment, so we can avoid the risk of shipping 2.2 without the BASICS ;)
Regards, Simone
Hi, I would like for ndiswrapper to appear on the iso, and perhaps madwifi. They make as much sense as having wireless-tools there. (That's there, right?) They're all relatively small. I think having these available could eliminate a bottleneck in the install process for wireless/laptop users (i.e. digging a CAT5 cable out of the closet ;)). Personally, the differentiation between opt and core hardly matters to me. To be frank, I think it's practically pointless. Core to me is not core to you which is not core to the next guy. And I don't mean for that remark to sound negative at all - an easily tailored system is something we all pursue and value. I look forward to testing 2.2 when my melted eMachines finds its way back to me. -- Jay Dolan jdolan.dyndns.org A: Top posting. Q: What's the most annoying thing about usenet? A: Because it's annoying to read. Q: Why is top-posting bad?
Simone Rota [2006-02-18 15:56]:
On 02/18/06 15:29 Juergen Daubert wrote:
To be honest, I'm not completely happy with the current divison of core/opt on the cd. I'd suggest the following changes: [...]
Since we're on topic, any other suggestion regarding the port to include/remove? In particular:
- Are we ok with the core/opt separation?
We recently discussed this and IIRC the majority of us voted for moving quite some stuff from the old opt to the new opt. I fail to understand why your development of 2.2 isn't taking place in Subversion, or why 2.2 hasn't been branched off yet so all of us can contribute to the release. Regards, Tilman -- A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing on usenet and in e-mail?
On 02/19/06 18:20 Tilman Sauerbeck wrote: Hi,
I fail to understand why your development of 2.2 isn't taking place in Subversion, or why 2.2 hasn't been branched off yet so all of us can contribute to the release.
The main reason for this is that I'd first like to hear more from other developers regarding the proposal of git migration (see Johannes' Feb 18 email), so we can eventually start working on a 2.2 git tree instead of using a short-lived 2.2 svn branch/repo. Considering that discussion on this topic is planned for this wednesday, I think it's ok to wait few days. Regards, Simone -- Simone Rota WEB : http://www.varlock.com Bergamo, Italy MAIL: sip@varlock.com
Simone Rota [2006-02-19 18:40]:
On 02/19/06 18:20 Tilman Sauerbeck wrote:
I fail to understand why your development of 2.2 isn't taking place in Subversion, or why 2.2 hasn't been branched off yet so all of us can contribute to the release.
The main reason for this is that I'd first like to hear more from other developers regarding the proposal of git migration (see Johannes' Feb 18 email), so we can eventually start working on a 2.2 git tree instead of using a short-lived 2.2 svn branch/repo.
Yeah, makes sense. Regards, Tilman -- A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing on usenet and in e-mail?
Tilman Sauerbeck wrote:
We recently discussed this and IIRC the majority of us voted for moving quite some stuff from the old opt to the new opt.
I fail to understand why your development of 2.2 isn't taking place in Subversion, or why 2.2 hasn't been branched off yet so all of us can contribute to the release. Hi,
I think Sip is refraining from putting his 2.2 changes into svn because he feels that it might "step on Per's toes." However, I agree that with a branch this would not be an issue. Also, I stole your signature :) I run a Quake2 mailing list for a bunch of goofballs who constantly top-post HTML replies from Outlook. It makes me sick. -- Jay Dolan jdolan.dyndns.org A: Top posting. Q: What's the most annoying thing about usenet? A: Because it's annoying to read. Q: Why is top-posting bad?
Jay Dolan wrote:
I think Sip is refraining from putting his 2.2 changes into svn because he feels that it might "step on Per's toes." However, I agree that with a branch this would not be an issue. Sorry. I wasn't trying to speak for [or over] Simone. His mail hadn't come through to me yet for some reason..
-- Jay Dolan jdolan.dyndns.org A: Top posting. Q: What's the most annoying thing about usenet? A: Because it's annoying to read. Q: Why is top-posting bad?
Hi, On Sun, Feb 19, 2006 at 18:20:15 +0100, Tilman Sauerbeck wrote: [...]
I fail to understand why your development of 2.2 isn't taking place in Subversion, or why 2.2 hasn't been branched off yet so all of us can contribute to the release. Basically because no one has created a branch so far and said "go now" :-).
In addition, toplevel copies can only be done by the admin group (which currently is Per, Matt and I), both to avoid unwanted commits and to keep branching under control; I wasn't aware that Simone was working on 2.2-devtest, otherwise I would have made a branch (I didn't want to create a branch back in january when I sent the mail regarding 2.2 since I really wanted to discuss it first, i.e. not rushing into it; later on when I probably should have done it I was kept busy IRL). Now, as Simone suggested I'd suggest to wait and see what we discuss tomorrow regarding Subversion; if we're agree to change it, we can start working on the branch immediately. Best regards, Johannes -- Johannes Winkelmann mailto:jw@tks6.net Zurich, Switzerland http://jw.tks6.net
Errr... On Tue, Feb 21, 2006 at 15:16:40 +0100, Johannes Winkelmann wrote: [...]
Now, as Simone suggested I'd suggest to wait and see what we discuss tomorrow regarding Subversion; if we're agree to change it, we can start working on the branch immediately. That should of course read "if we agree to keep it"; even though it doesn't really matter, we can start working on 2.2 either way ;-).
Best regards, Johannes -- Johannes Winkelmann mailto:jw@tks6.net Zurich, Switzerland http://jw.tks6.net
Hi, make boostrap failed at stage 1, on grub. The odd thing, is that it build fine during stage 0. Attached are the build logs. Cheers, Nick P.S. clc-devel moderator: sorry about forgetting to bzip2 those attachments, when I first sent this.
On 02/25/06 12:12 Nick Steeves wrote:
Hi,
Hi,
make boostrap failed at stage 1, on grub. The odd thing, is that it build fine during stage 0. Attached are the build logs.
You may have some old toolchain leftover in stage0 binaries that make stage1 compilation fail. In this case the error should go away after a second bootstrap try. Of course it's also possible I messed things up with the grub patches; I'm currently working on devtest3, I'll double-check the grub port. Thanks for reporting, Regards, Simone -- Simone Rota WEB : http://www.varlock.com Bergamo, Italy MAIL: sip@varlock.com
On Saturday 25 February 2006 5:05, Simone Rota wrote:
You may have some old toolchain leftover in stage0 binaries that make stage1 compilation fail. In this case the error should go away after a second bootstrap try. Attached is my 2nd try build output--just another "make bootstrap", after removing the old log files.
Cheers, Nick
On Saturday 25 February 2006 5:05, Simone Rota wrote:
Of course it's also possible I messed things up with the grub patches; I'm currently working on devtest3, I'll double-check the grub port.
Since we have such old linux-libc-headers (specifically pre-2.6.13), is it possible for us to add the following to our glibc package for devtest3?: http://www.kernel.org/pub/linux/kernel/people/rml/inotify/headers/inotify-sy... http://www.kernel.org/pub/linux/kernel/people/rml/inotify/headers/inotify.h These headers are very much desktop-related. Currently, AFAIK, only Beagle (in the gnome repo), Gamin, and KAT (in my personal repo) benefit from it. Ah yes, XFCE (http://xfce-diary.blogspot.com/2005_08_01_xfce-diary_archive.html) via Thunar via Gamin, KDE via Gamin, and Enlightenment-0.17 (http://www.enlightenment.org/Main/News/_articles/8.html) also use, or will use it. Inotify is the replacement for DNotify, etc., etc., and it would be nice if we had support for it in CRUX-2.2, seeing as the major desktop environments are moving towards it. Cheers, Nick
Hi, I looked into those inotify headers a bit. They're not in sync with linux-2.6.15.x, so I replaced inotify.h with one from linux-2.6.15.5. Inotify-syscalls.h only seems to be necessary if one's glibc hasn't been compiled against inotify.h. What I wonder is whether or not glibc-2.3.6 can, and will build the syscalls bit merely from adding inotify.h to $PKG/usr/include/linux? I'm currently swamped with homework, so won't be able to upgrade my kernel (which will probably involve wrestling with the at76c503c driver...yay) and glibc until the weekend (hopefully). Cheers, Nick
According to (http://www.diy-linux.org/pipermail/diy-linux-dev/2005-October.txt), inotify support in glibc is only in the 2.4 branch. Hmm. I wonder how to use that syscalls header? I read somewhere that it can be >>'d to inotify.h, but am not sure how well that will work.
Hi, On Thu, Mar 02, 2006 at 02:02:20 -0700, Nick Steeves wrote:
According to (http://www.diy-linux.org/pipermail/diy-linux-dev/2005-October.txt), inotify support in glibc is only in the 2.4 branch. Hmm. I wonder how to use that syscalls header? I read somewhere that it can be >>'d to inotify.h, but am not sure how well that will work. Maybe we could have a look at opensuse/novell's glibc, after all Robert Love works there.
Maybe we can just use http://www.kernel.org/pub/linux/kernel/people/rml/inotify/headers/ ? Regards, Johannes -- Johannes Winkelmann mailto:jw@tks6.net Zurich, Switzerland http://jw.tks6.net
On Thursday 02 March 2006 2:17, Johannes Winkelmann wrote:
Maybe we could have a look at opensuse/novell's glibc, after all Robert Love works there. Hmm, that's a good idea.
Maybe we can just use http://www.kernel.org/pub/linux/kernel/people/rml/inotify/headers/ That was my original intent, but then I noticed that the that inotify.h is quite a bit different then the one in my kernel (2.6.14.2), and in what will probably be the kernel most CRUX-2.2 users will use (2.6.15.5+). I've attached diffs for each to illustrate the changes.
Update: Well, I copied the inotify.h from linux-2.6.15.5/include/linux/inotify.h to /usr/include/linux/inotify.h, rebuilt gamin with debugging enabled, and it finds and uses inotify (where before it used dnotify). Using the inotify.h from 2.6.14.2 results in lots of "wd" errors. Does "wd" stand for word-descriptors, and are this file-system related? At any rate, the inotify.h from linux-2.6.15.5 works great. Attached are diffs for showing how the different inotify.h's are different, and my glibc/Pkgfile patch, and linux-2.6.15.5/include/linux/inotify.h. Cheers, Nick
Oops, I clicked send instead of attach.
participants (6)
-
Jay Dolan -
Johannes Winkelmann -
Juergen Daubert -
Nick Steeves -
Simone Rota -
Tilman Sauerbeck