Bug#333001: mpich: FTBFS on GNU/kFreeBSD
On Fri, 2006-05-26 at 11:56 +0200, Petr Salinger wrote: > Hi, > > please could you support mpich and dependents on kFreeBSD. > Attached please find a modified and extended patch, > updating of config.{guess,sub} is done at build time, > not via dpatch. This variant is easily reviewable. Hello. Thank you for this variant, I really like it much better than the old one. I will upload my next version with it included, and you will be notified that the bug has been closed; please test it to make sure it builds for you. Apologies for the long delay in getting to this! Regards, -Adam -- GPG fingerprint: D54D 1AEE B11C CE9B A02B C5DD 526F 01E8 564E E4B6 Welcome to the best software in the world today cafe! http://www.take6.com/albums/greatesthits.html -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#333001: mpich: FTBFS on GNU/kFreeBSD
Hi, please could you support mpich and dependents on kFreeBSD. Attached please find a modified and extended patch, updating of config.{guess,sub} is done at build time, not via dpatch. This variant is easily reviewable. Thanks in advance. Petr diff -u mpich-1.2.7/debian/control mpich-1.2.7/debian/control --- mpich-1.2.7/debian/control +++ mpich-1.2.7/debian/control @@ -3,7 +3,7 @@ Priority: extra Maintainer: Adam C. Powell, IV <[EMAIL PROTECTED]> Standards-Version: 3.6.2.1 -Build-Depends: g77, rsh-client, libx11-dev, libxt-dev, file, debhelper (>> 4.0.0), dpatch, tk8.4 +Build-Depends: g77, rsh-client, libx11-dev, libxt-dev, file, debhelper (>> 4.0.0), dpatch, tk8.4, autotools-dev Build-Depends-Indep: bzip2 Package: mpich-bin diff -u mpich-1.2.7/debian/rules mpich-1.2.7/debian/rules --- mpich-1.2.7/debian/rules +++ mpich-1.2.7/debian/rules @@ -17,6 +17,9 @@ CONFIGURE_OPTIONS= -rsh=/usr/bin/rsh +CONFIG_SG_DIRS= mpe mpe/slog2sdk mpe/slog2sdk/trace_rlog mpe/slog2sdk/trace_sample \ + examples/perftest/config/confdb src/fortran/config MPI-2-C++ romio/confdb + # checks the environmental variable DEB_BUILD_OPTIONS ifneq (,$(findstring debug,$(DEB_BUILD_OPTIONS))) CONFIGURE_OPTIONS+= --enable-devdebug --enable-debug @@ -46,6 +49,9 @@ build: build-stamp build-stamp: patch-stamp dh_testdir + for CFGAUXDIR in $(CONFIG_SG_DIRS) ; do \ + cp -f /usr/share/misc/config.{sub,guess} $$CFGAUXDIR ; \ + done # make mpich normal (test -d image_mpich || mkdir image_mpich; cd image_mpich && ../configure \ $(CONFIGURE_OPTIONS_P4) && \ @@ -79,6 +85,9 @@ clean: unpatch dh_testdir dh_testroot + for CFGAUXDIR in $(CONFIG_SG_DIRS) ; do \ + rm -f $$CFGAUXDIR/config.{sub,guess} ; \ + done rm -f build-stamp install-stamp # clean up the build directory (set -e; for INST in mpich mpich-mpd mpich-shmem ; do \ diff -u mpich-1.2.7/debian/patches/00list mpich-1.2.7/debian/patches/00list --- mpich-1.2.7/debian/patches/00list +++ mpich-1.2.7/debian/patches/00list @@ -14,0 +15 @@ +19_kfreebsd only in patch2: unchanged: --- mpich-1.2.7.orig/debian/patches/19_kfreebsd.dpatch +++ mpich-1.2.7/debian/patches/19_kfreebsd.dpatch @@ -0,0 +1,42 @@ +#! /bin/sh -e +## 19_kfreebsd.dpatch by Aurelien Janro <[EMAIL PROTECTED]> +## +## All lines beginning with `## DP:' are a description of the patch. +## DP: Add support for GNU/kFreeBSD + +if [ $# -ne 1 ]; then +echo >&2 "`basename $0`: script expects -patch|-unpatch as argument" +exit 1 +fi +case "$1" in +-patch) patch -f --no-backup-if-mismatch -p1 < $0;; +-unpatch) patch -f --no-backup-if-mismatch -R -p1 < $0;; +*) +echo >&2 "`basename $0`: script expects -patch|-unpatch as argument" +exit 1;; +esac + +exit 0 + +--- mpich-1.2.7.orig/bin/tarch mpich-1.2.7/bin/tarch +@@ -138,6 +138,7 @@ + next)FARCH=NeXT ; break ;; + KSR1|KSR2) FARCH=ksr ; break ;; + FreeBSD) FARCH=freebsd ; break ;; ++GNU/kFreeBSD) FARCH=freebsd ; break ;; + NetBSD) FARCH=netbsd ; break ;; + i386)GARCH=ipsc2 ;; + ULTRIX|RISC) GARCH=dec5000 ;; + +--- mpich-1.2.7.orig/mpid/ch_p4/p4/lib/p4_secure.c mpich-1.2.7/mpid/ch_p4/p4/lib/p4_secure.c +@@ -367,6 +367,7 @@ + #if defined(P4BSD) && !defined(NO_ECHO) + + #ifdef FREEBSD ++#include + #include + #else + #include +
Bug#333001: mpich: FTBFS on GNU/kFreeBSD
On Sun, 15 Jan 2006, Adam C Powell IV wrote: > On Sun, 2006-01-15 at 23:23 +0100, Santiago Vila wrote: > > On Sun, 9 Oct 2005, Adam C Powell IV wrote: > > > > > It looks like about 80-90% of the patch is unnecessary to support this > > > architecture, e.g. there's no need for openrisc, os400, djgpp, crx etc. > > > in a Debian system. Can you please edit out the unnecessary parts so > > > the patch is more compact, while supporting the new arch (and other new > > > ones like armeb)? If you don't, I will try, but I can't guarantee it > > > will work... > > > > Well, there might be no need for openrisc, os400, djgpp, crx, but if > > you think about it, there is really no need to make the patch shorter > > either. > > Sure there is. It's about conciseness and verifiability. I want a set > of small patches which someone can reasonably say, "Okay, I see what > that does, it does that and nothing more," and not worry that it will do > something bad to their machine. Long patches which do much more than is > needed for Debian are inherently bad. I understand that you prefer a small patch over a big patch *in general*, but a small patch is not always more verifiable. In this case, for example, if you use the most recent config.guess and config.guess, it will be easy to verify that the modified versions match the ones in the autotools-dev package. The latest versions of those files in the autotools-dev package are the ones we *fully* trust, for *every* machine. Everybody can reasonably say "okay, I see that the patch updates those files to the latest version, which I trust". You even admit that if you try to make the patch smaller, you don't guarantee that it will work. We know that the full config.guess and config.sub files do work and we trust them. Why lose your time (or ask anybody else to lose his time) trying to make the patch smaller when doing so increases the risk of breaking something? I for one would certainly not consider this package "safer" just because you managed to save a few bytes in the size of the patch. In such case I would start worrying about it, not the opposite. So I agree that a small patch is better than a long patch in general, but there are exceptions, and this is one of them. If you try to make it smaller, the probability of breaking something increases, not decreases. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#333001: mpich: FTBFS on GNU/kFreeBSD
On Sun, 2006-01-15 at 23:23 +0100, Santiago Vila wrote: > On Sun, 9 Oct 2005, Adam C Powell IV wrote: > > > It looks like about 80-90% of the patch is unnecessary to support this > > architecture, e.g. there's no need for openrisc, os400, djgpp, crx etc. > > in a Debian system. Can you please edit out the unnecessary parts so > > the patch is more compact, while supporting the new arch (and other new > > ones like armeb)? If you don't, I will try, but I can't guarantee it > > will work... > > Well, there might be no need for openrisc, os400, djgpp, crx, but if > you think about it, there is really no need to make the patch shorter > either. Sure there is. It's about conciseness and verifiability. I want a set of small patches which someone can reasonably say, "Okay, I see what that does, it does that and nothing more," and not worry that it will do something bad to their machine. Long patches which do much more than is needed for Debian are inherently bad. > In fact, this is not really about applying a patch, it is about using > the most recent config.guess and config.sub available. You can find > those files in the autotools-dev package, in /usr/share/misc. Right, but that's an issue for upstream, which has stopped developing mpich-1.x. > If you don't like adding a patch which is aesthetically ugly to you, > you can copy the files at build time instead. The details are > explained in /usr/share/doc/autotools-dev/README.Debian.gz. Interesting idea. At some point I might try this, but for now I think I'll just cut a lot out of your patch. Thanks for the suggestion. > > Also, mpich is stuck with about twelve other packages trying to enter > > testing, so until that transition is made I'll be unable to upload a new > > mpich package. I'll do this as soon as possible. > > I hope such is not the case anymore. Nope, so I'll try to get to this soon. > For the record, several hurd-i386 packages fail to build from source > because libmpich1.0-dev does not exist. Yes, mpich is pretty central to much of the cluster computing infrastructure. I'll try to get to it soon. Thanks, -Adam -- GPG fingerprint: D54D 1AEE B11C CE9B A02B C5DD 526F 01E8 564E E4B6 Welcome to the best software in the world today cafe! http://www.take6.com/albums/greatesthits.html -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#333001: mpich: FTBFS on GNU/kFreeBSD
On Sun, 9 Oct 2005, Adam C Powell IV wrote: > It looks like about 80-90% of the patch is unnecessary to support this > architecture, e.g. there's no need for openrisc, os400, djgpp, crx etc. > in a Debian system. Can you please edit out the unnecessary parts so > the patch is more compact, while supporting the new arch (and other new > ones like armeb)? If you don't, I will try, but I can't guarantee it > will work... Well, there might be no need for openrisc, os400, djgpp, crx, but if you think about it, there is really no need to make the patch shorter either. In fact, this is not really about applying a patch, it is about using the most recent config.guess and config.sub available. You can find those files in the autotools-dev package, in /usr/share/misc. If you don't like adding a patch which is aesthetically ugly to you, you can copy the files at build time instead. The details are explained in /usr/share/doc/autotools-dev/README.Debian.gz. > Also, mpich is stuck with about twelve other packages trying to enter > testing, so until that transition is made I'll be unable to upload a new > mpich package. I'll do this as soon as possible. I hope such is not the case anymore. For the record, several hurd-i386 packages fail to build from source because libmpich1.0-dev does not exist. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#333001: mpich: FTBFS on GNU/kFreeBSD
On Mon, 2005-10-10 at 00:37 +0200, Aurelien Jarno wrote: > Package: mpich > Severity: important > Tags: patch > > Hi, > > mpich fails to build on GNU/kFreeBSD due to missing support and outdated > config.{guess,sub}. Please find attached a patch to fix that. Could you > please add it in the next upload? > > Don't hesitate to contact me if you need more information. > > Thanks in advance, > Aurelien Thank you for the bug, and patch. I appreciate the work you are doing to support Debian on this new kernel/OS combination. It looks like about 80-90% of the patch is unnecessary to support this architecture, e.g. there's no need for openrisc, os400, djgpp, crx etc. in a Debian system. Can you please edit out the unnecessary parts so the patch is more compact, while supporting the new arch (and other new ones like armeb)? If you don't, I will try, but I can't guarantee it will work... Also, mpich is stuck with about twelve other packages trying to enter testing, so until that transition is made I'll be unable to upload a new mpich package. I'll do this as soon as possible. Thanks again, -Adam -- GPG fingerprint: D54D 1AEE B11C CE9B A02B C5DD 526F 01E8 564E E4B6 Welcome to the best software in the world today cafe! http://www.take6.com/albums/greatesthits.html -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]