- Arch-dev-public - lists.archlinux.org

Re: [arch-dev-public] [arch-general] commitpkg script than handle better splitted package
by Dan McGee 18 Aug '10

18 Aug '10

On Tue, Aug 17, 2010 at 5:12 AM, Laurent Carlier <lordheavym(a)gmail.com> wrote: > Le lundi 16 août 2010 17:30:54, vous avez écrit : >> Le lundi 16 août 2010 17:24:29, vous avez écrit : >> > Le lundi 16 août 2010 14:10:10, Dan McGee a écrit : >> > > Patches are a lot more likely to get looked at by any of us. And yes, >> > > this should probably be more than one patch. >> > > >> > > http://projects.archlinux.org/devtools.git/ >> > >> > Hre are the patches (5), i hope i've not introduce bug while merging my >> > changes to the git repo :-p >> > >> > 0001-Moved-file-s-sending-to-the-staging-dir-to-the-funct.patch >> > 0002-Doesn-t-manage-monolithic-and-splitted-packages-the-.patch >> > 0003-Detect-if-arch-is-redefined-in-the-package-function-.patch >> > 0004-Now-manage-if-arch-is-redefined-and-not-in-default-a.patch >> > 0005-Manage-redefintion-of-pkgver-and-pkgver-in-the-packa.patch >> >> patches are available here: >> http://pkgbuild.com/~lcarlier/ >> >> Regards, > > Rebased the patches, and add another patch: > 0006-Remove-useless-number-from-continue.patch Pierre, is this something you can review since you have taken the lead on devtools/dbscripts? -Dan

2 1

Re: [arch-dev-public] [arch-dev] [pacman-dev] [PATCH] repo-add: create relative DB symlinks
by Dan McGee 18 Aug '10

18 Aug '10

On Wed, Aug 18, 2010 at 1:57 PM, Pierre Schmitz <pierre(a)archlinux.de> wrote: > On Wed, 18 Aug 2010 13:23:16 -0500, Dan McGee <dan(a)archlinux.org> > wrote: >> As noted in FS#20498, if an absolute path is used for specifying the >> database when invoking repo-add, the symlink generated will point to the >> absolute path instead of being relative to the directory. Fix this for >> the two linking cases, but leave the copy untouched so that will still >> work. >> >> Signed-off-by: Dan McGee <dan(a)archlinux.org> >> --- >> scripts/repo-add.sh.in | 5 +++-- >> 1 files changed, 3 insertions(+), 2 deletions(-) >> >> diff --git a/scripts/repo-add.sh.in b/scripts/repo-add.sh.in >> index 2838f81..283ad07 100644 >> --- a/scripts/repo-add.sh.in >> +++ b/scripts/repo-add.sh.in >> @@ -488,8 +488,9 @@ if (( success )); then >> [[ -f $REPO_DB_FILE ]] && mv -f "$REPO_DB_FILE" "${REPO_DB_FILE}.old" >> [[ -f $tmpdir/$filename ]] && mv "$tmpdir/$filename" "$REPO_DB_FILE" >> dblink="${REPO_DB_FILE%.tar.*}" >> - ln -sf "$REPO_DB_FILE" "$dblink" 2>/dev/null || \ >> - ln -f "$REPO_DB_FILE" "$dblink" 2>/dev/null || \ >> + target=${REPO_DB_FILE##*/} >> + ln -sf "$target" "$dblink" 2>/dev/null || \ >> + ln -f "$target" "$dblink" 2>/dev/null || \ >> cp "$REPO_DB_FILE" "$dblink" >> else >> msg "$(gettext "No packages modified, nothing to do.")" > > Maybe we could already apply this on gerolde. ATM we are pushing those > broken links to the mirrors. (yes, this didn't happen with the old > dbscripts) No need for the private list. And go ahead if you want, but since no program even uses these links yet, it really isn't a rush at all. It is also going in 3.4.1 which Allan and I want to release in the next week or so. -Dan

2 1

Re: [arch-dev-public] [arch-general] Mailman update
by Allan McRae 18 Aug '10

18 Aug '10

On 24/07/10 02:55, Dan Vratil wrote: > I tried the latest version of mailman (2.1.13) as you suggested. I used the > PKGBUILD from SVN trunk, but I modified it, so that it does not require > /home/mailman and mailman user and group to exist during build and I modified > the mailman.install file so that useradd does not create /home/mailman. > I'm not sure if this ML allows attachments, so the source tarball is on > http://pub.progdan.cz/mailman-2.1.13-1.src.tar.gz if any dev were interested. > > The mailman is working without problems, but making it work was not as simple > as I thought at the beginning :) Our recent wiki page about mailman is > absolutely useless. I would like to update it, put there a guide how to set up > mailman (with postfix), but it's pointless until the package is updated. I have looked at the PKGBUILD and quickly built it. Everything looks good as far as I can tell. The install file still has the old syntax to be removed. Can someone do the needed builds and upload this update to [testing]? Allan

3 5

[arch-dev-public] Large packages in repositories
by Dan McGee 17 Aug '10

17 Aug '10

Hey guys, A package went in so big today that it made reporead blow up on my local database due to the installed size being > 2GB: http://www.archlinux.org/packages/community/i686/sage-mathematics/ http://www.archlinux.org/packages/community/x86_64/sage-mathematics/ File "/usr/lib/python2.6/site-packages/django/db/backends/postgresql_psycopg2/base.py", line 44, in execute return self.cursor.execute(query, args) django.db.utils.DatabaseError: integer out of range I'm wondering if we need to be more careful when it comes to these big packages entering our repositories. This one is especially suspect as of its 71096 files (and 71094 in the other architecture), a ton of them are things like *.py, *.pyc, *.html, or *.png files. This is ripe for splitting into a -data package (or not including some of this junk, if it is that, at all). mysql> select count(*), substring_index(path, '.', -1) from package_files where pkg_id = 49860 and path like '%.%' group by substring_index(path, '.', -1) order by count(*) desc limit 25; +----------+--------------------------------+ | count(*) | substring_index(path, '.', -1) | +----------+--------------------------------+ | 29787 | png | | 9410 | py | | 6602 | pyc | | 1804 | h | | 1722 | html | | 1659 | txt | | 1168 | pyo | | 988 | doctree | | 988 | rst | | 886 | hpp | tl;dr: I think we need some standards with these huge packages, and people need to be a lot more cognizant as to how big they are. We have lost more than one mirror due to complaints over needed space and stuff like this doesn't help. If any TU's would like to forward this along and solicit their thoughts I'd appreciate it. -Dan

3 5

[arch-dev-public] [PATCH] copy packages into TEMPDIR
by Ionut Biru 17 Aug '10

17 Aug '10

TEMPDIR is used before packages are copied into it and made it fail because cannot find any. Signed-off-by: Ionut Biru <ibiru(a)archlinux.org> --- checkpkg | 13 +++++++------ 1 files changed, 7 insertions(+), 6 deletions(-) diff --git a/checkpkg b/checkpkg index 45572ca..0987a85 100755 --- a/checkpkg +++ b/checkpkg @@ -27,6 +27,7 @@ if [ "$arch" == 'any' ]; then CARCH='any' fi +STARTDIR=$(pwd) TEMPDIR=$(mktemp -d /tmp/checkpkg-script.XXXX) cd $TEMPDIR @@ -34,14 +35,14 @@ for _pkgname in ${pkgname[@]}; do pkgfile=${_pkgname}-${pkgver}-${pkgrel}-${CARCH}${PKGEXT} oldstylepkgfile=${_pkgname}-${pkgver}-${pkgrel}${PKGEXT} - if [ -f "$(pwd)/$pkgfile" ]; then - pkgfile=$(pwd)/$pkgfile + if [ -f "$STARTDIR/$pkgfile" ]; then + cp "$STARTDIR/$pkgfile" . elif [ -f "$PKGDEST/$pkgfile" ]; then - pkgfile=$PKGDEST/$pkgfile - elif [ -f "$(pwd)/$oldstylepkgfile" ]; then - pkgfile=$(pwd)/$oldstylepkgfile + cp "$PKGDEST/$pkgfile" . + elif [ -f "$STARTDIR/$oldstylepkgfile" ]; then + cp "$STARTDIR/$oldstylepkgfile" . elif [ -f "$PKGDEST/$oldstylepkgfile" ]; then - pkgfile=$PKGDEST/$oldstylepkgfile + cp "$PKGDEST/$oldstylepkgfile" . else echo "File \"$pkgfile\" doesn't exist" exit 1 -- 1.7.2.1

1 0

[arch-dev-public] [PATCH] minor fix to error message
by Allan McRae 17 Aug '10

17 Aug '10

Signed-off-by: Allan McRae <allan(a)archlinux.org> --- db-update | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/db-update b/db-update index b11face..51e3cc2 100755 --- a/db-update +++ b/db-update @@ -36,7 +36,7 @@ for repo in ${repos[@]}; do die "Package ${repo}/$(basename ${pkg}) is not consistent with svn repository" fi if [ -f "${FTP_BASE}/$(get_pkgpool_for_host)/$(basename ${pkg})" ]; then - die "Package ${repo}/$(basename pkg) already exists in ${FTP_BASE}/$(get_pkgpool_for_host)" + die "Package ${repo}/$(basename ${pkg}) already exists in ${FTP_BASE}/$(get_pkgpool_for_host)" fi done else -- 1.7.2.1

1 0

[arch-dev-public] Fwd: [arch-notifications] Core/Extra Cleanup 17-08-2010
by Eric Bélanger 17 Aug '10

17 Aug '10

==> The file "/srv/ftp/staging/os/i686/staging.db.tar.gz" could not be found, skipping. ==> The file "/srv/ftp/staging/os/x86_64/staging.db.tar.gz" could not be found, skipping. Should we add empty db (tar cvzf staging.db.tar.gz -T /dev/null) ?

2 1

[arch-dev-public] [signoff] kernel26 2.6.35.2-1
by Tobias Powalowski 16 Aug '10

16 Aug '10

Latest kernel is in testing, please signoff for both arches. greetings tpowa -- Tobias Powalowski Archlinux Developer & Package Maintainer (tpowa) http://www.archlinux.org tpowa(a)archlinux.org

4 6

[arch-dev-public] /dev/shm permissions and building in chroot
by Allan McRae 16 Aug '10

16 Aug '10

Hi, I have been tracking down a python3 bug and it turns out to be caused by our chroot building. Essentially the permissions of /dev/shm are different in the chroot than on the system: allan@mugen /home/arch/chroot/stable-i686/copy/dev > ls -ld shm drwxr-xr-x 2 root root 40 Jul 12 15:16 shm allan@mugen /dev > ls -ld shm drwxrwxrwt 2 root root 40 Aug 14 10:23 shm So when configure scripts try to test for POSIX semaphores by writing in that directory, they fail. Looking at the relevant line in mkarchroot: mount -o bind /dev "${working_dir}/dev" Doing that manually show the same issue. Should we add chmod 1777 "${working_dir}/dev/shm" after the mount? It fixes my issue but feels hackish. Any other suggestions? Allan

2 3

[arch-dev-public] [signoff] syslog-ng-3.1.2-1
by Allan McRae 15 Aug '10

15 Aug '10

Upstream changelog: INCOMPATIBLE CHANGES: * This release changes the default for 'store-legacy-msghdr' flag for log sources. In previous releases this had to be enabled explicitly, because of the performance penalties, which do not apply to 3.1. The original behaviour can be restored by explicitly specifying 'dont-store-legacy-msghdr'. Bugfixes: * When syslog-ng is reloaded, the local hostname value was not refreshed, causing syslog-ng to remember the hostname until the next restart. This may not play nice with DHCP configured hostnames, which may change dynamically. * When several SIGHUP signals are received in quick succession, the last one may have been dropped. This problem is fixed. * Make sure that numbers never get resolved using getpwnam/getgrnam (e.g. libnss functions), because this may cause deadlocks if the NSS provider is LDAP and the LDAP server is trying to log a message to syslog about invalid usernames. * Fixed flush_timeout() handling in for non-file destinations. Enabling flush_lines()/flush_timeout() for a non-file destination, either using a global option, or a per-destination option could cause excess CPU usage as long as the destination wasn't writable. * In case patterndb has a syntax error when reloading the file, syslog-ng automatically reverts to the old version instead of dropping pattern matching altogether. * Fixed pdbtool to properly handle accented characters in the command line. * Fixed the use of the greedy and drop-invalid flags for csv-parser. * Fixed a possible but rare memory leak in the handling of message suppression. Other changes: * Added "update-patterndb" command that merges patterndb files from ${sysconfdir}/patterndb.d and produces ${localstatedir}/patterndb.xml, the default file of the db-parser(). * Make it possible to specify the path to the syslog-ng control socket using a command line option in order to make it easier to launch multiple syslog-ng instances. * Introduced a new, cleaner syntax to specify that the permissions of a file are not to be changed. Use owner/group/perm options of a file without any parameters instead of "-1". The old mechanism had the problem that both per-file and global settings had to use "-1" to effectively stop syslog-ng from changing the uid/gid/perm values. The new syntax will make it possible to specify defaults in the global options section and customize those on a per-file basis. * syslog-ng will not go into the background if either the --syntax-only or --debug options are specified. Loggen: * Added SSL support. * Added the ability to read from a sample file Signoff both, Allan

3 2