[pacman-dev] 3.1 release, pacman-git builds, etc.
I'll try to keep this short and sweet, and hopefully this wakes up the ghosts on the mailing list and we can make a push to get 3.1 released sometime in the relatively soon future. I've made a PKGBUILD that I threw together tonight to actually test pacman on my system without killing the working pacman 3.0.X install. It installs pacman and all its files to /usr/local, which is rather convenient for testing as you can just switch that directory in and out of your path as necessary. It also skips the complicated renames of files, and as a bonus, gives our autoconf script some testing with alternate install locations. However, it is configured to use your current db and cache at the /var location, so don't think that is necessarily safe (although it better be). Here is the most recent build as of when I am writing this email, use at your own risk! http://www.archlinux.org/~dan/pacman-git-3.1.0devel-2-i686.pkg.tar.gz http://www.archlinux.org/~dan/pacman-git_PKGBUILD With that said, I need your guys help testing this stuff out AND fixing the bugs and snags we hit that prevent us from releasing. The things on my list that definitely need to be done: * Ensure our config file changes are not necessary; e.g. old config files should mostly work with the new software. I fixed a few of these issues tonight. * General usability tests. Make sure things are acting as they should, no regressions, etc. * Merge the asciidoc branch and get our manpages up to snuff. * Figure out why Doxygen is generating some extra unnecessary manpages, and find a way to kill it. * makepkg testing. A lot of it, I haven't really used the new one at all since I tend to build official packages with the current release of makepkg to make sure it doesn't break things. * Translation updates. This could be a killer, but even if people don't get all the scripts translated we should try to get the pacman/libalpm translations back up to 100%. We lost a lot of messages because we cut debug translations, so that should help. * Testing of every bug on Flyspray marked "requires testing" and set to be fixed in 3.1. * Anything else you guys can think of. Please reply and post it and make sure it gets addressed. I've already got some ideas for 3.2/4.0 up my sleeve, but I'll put those in another email so we stay on topic here. And unless your idea is absolutely essential for a 3.1 release, try to keep them there as well. -Dan
Dan McGee wrote:
I'll try to keep this short and sweet, and hopefully this wakes up the ghosts on the mailing list and we can make a push to get 3.1 released sometime in the relatively soon future.
I've made a PKGBUILD that I threw together tonight to actually test pacman on my system without killing the working pacman 3.0.X install. It installs pacman and all its files to /usr/local, which is rather convenient for testing as you can just switch that directory in and out of your path as necessary. It also skips the complicated renames of files, and as a bonus, gives our autoconf script some testing with alternate install locations. However, it is configured to use your current db and cache at the /var location, so don't think that is necessarily safe (although it better be).
Here is the most recent build as of when I am writing this email, use at your own risk! http://www.archlinux.org/~dan/pacman-git-3.1.0devel-2-i686.pkg.tar.gz http://www.archlinux.org/~dan/pacman-git_PKGBUILD
With that said, I need your guys help testing this stuff out AND fixing the bugs and snags we hit that prevent us from releasing. The things on my list that definitely need to be done:
* Ensure our config file changes are not necessary; e.g. old config files should mostly work with the new software. I fixed a few of these issues tonight. * General usability tests. Make sure things are acting as they should, no regressions, etc. * Merge the asciidoc branch and get our manpages up to snuff. * Figure out why Doxygen is generating some extra unnecessary manpages, and find a way to kill it. * makepkg testing. A lot of it, I haven't really used the new one at all since I tend to build official packages with the current release of makepkg to make sure it doesn't break things. * Translation updates. This could be a killer, but even if people don't get all the scripts translated we should try to get the pacman/libalpm translations back up to 100%. We lost a lot of messages because we cut debug translations, so that should help. * Testing of every bug on Flyspray marked "requires testing" and set to be fixed in 3.1. * Anything else you guys can think of. Please reply and post it and make sure it gets addressed.
I've already got some ideas for 3.2/4.0 up my sleeve, but I'll put those in another email so we stay on topic here. And unless your idea is absolutely essential for a 3.1 release, try to keep them there as well.
-Dan
_______________________________________________ pacman-dev mailing list pacman-dev@archlinux.org http://archlinux.org/mailman/listinfo/pacman-dev I've built quite a few packages with makepkg-3.1 and it appears to be working ok :)
Andrew
On 8/22/07, Andrew Fyfe <andrew@neptune-one.net> wrote:
I've built quite a few packages with makepkg-3.1 and it appears to be working ok :)
We don't have much of a choice but to tell people to merge their makepkg.conf file, do we? It looks like we have changed the structure a good amount from before with the addition of DLAGENT and all the variables at the bottom as well. -Dan
2007/8/22, Dan McGee <dpmcgee@gmail.com>:
On 8/22/07, Andrew Fyfe <andrew@neptune-one.net> wrote:
I've built quite a few packages with makepkg-3.1 and it appears to be working ok :)
We don't have much of a choice but to tell people to merge their makepkg.conf file, do we? It looks like we have changed the structure a good amount from before with the addition of DLAGENT and all the variables at the bottom as well.
Seems like "OK" solution to me. It's not the first time Arch users have to adapt to new config (not in pacman only). -- Roman Kyrylych (Роман Кирилич)
On Wed, Aug 22, 2007 at 06:13:16AM +0100, Andrew Fyfe wrote:
I've built quite a few packages with makepkg-3.1 and it appears to be working ok :)
I thought it worked since I first tried it on pacman-git PKGBUILD. But it actually fails with most PKGBUILDs using source tarballs.
From 223d5606cab31c8695b9516ba21434da5529dedf Mon Sep 17 00:00:00 2001 From: Chantry Xavier <shiningxc@gmail.com> Date: Thu, 23 Aug 2007 20:00:41 +0200 Subject: [PATCH] makepkg : add -L (--dereference) to file.
makepkg didn't correctly detect the type of the archive since it didn't dereference symlinks, and so failed to extract the source tarball. Signed-off-by: Chantry Xavier <shiningxc@gmail.com> --- scripts/makepkg.sh.in | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in index eb42cef..e9578b2 100644 --- a/scripts/makepkg.sh.in +++ b/scripts/makepkg.sh.in @@ -611,7 +611,7 @@ extract_sources() { fi # fix flyspray #6246 - local file_type=$(file -biz "$file") + local file_type=$(file -bizL "$file") local cmd='' case "$file_type" in *application/x-tar*|*application/x-zip*|*application/x-cpio*) -- 1.5.2.5
Xavier wrote:
On Wed, Aug 22, 2007 at 06:13:16AM +0100, Andrew Fyfe wrote:
I've built quite a few packages with makepkg-3.1 and it appears to be working ok :)
I thought it worked since I first tried it on pacman-git PKGBUILD. But it actually fails with most PKGBUILDs using source tarballs.
From 223d5606cab31c8695b9516ba21434da5529dedf Mon Sep 17 00:00:00 2001 From: Chantry Xavier <shiningxc@gmail.com> Date: Thu, 23 Aug 2007 20:00:41 +0200 Subject: [PATCH] makepkg : add -L (--dereference) to file.
makepkg didn't correctly detect the type of the archive since it didn't dereference symlinks, and so failed to extract the source tarball.
Signed-off-by: Chantry Xavier <shiningxc@gmail.com> --- scripts/makepkg.sh.in | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in index eb42cef..e9578b2 100644 --- a/scripts/makepkg.sh.in +++ b/scripts/makepkg.sh.in @@ -611,7 +611,7 @@ extract_sources() { fi
# fix flyspray #6246 - local file_type=$(file -biz "$file") + local file_type=$(file -bizL "$file") local cmd='' case "$file_type" in *application/x-tar*|*application/x-zip*|*application/x-cpio*) Oops I forgot to send that patch :) I've got another patch I forgot to send as well... http://neptune-one.homeip.net/git?p=pacman;a=commit;h=ea70e23cbbf17e5f00b564...
Andrew
On Sat, Aug 25, 2007 at 05:03:29PM +0100, Andrew Fyfe wrote:
Oops I forgot to send that patch :) I've got another patch I forgot to send as well... http://neptune-one.homeip.net/git?p=pacman;a=commit;h=ea70e23cbbf17e5f00b564...
Oh I see, I found it strange it worked fine for you :) I made this second patch as well :) http://chantry.homelinux.org/~xav/git/gitweb.cgi?p=pacman.git;a=commitdiff;h... Didn't have other issues though.
After installing filesystem, here is what I get in the database : %BACKUP% etc/fstab 0d3cfed6eacf21072e468bb7fe80cf07 etc/crypttab etc/group etc/hosts etc/ld.so.conf etc/passwd etc/shadow etc/gshadow etc/resolv.conf etc/motd etc/nsswitch.conf etc/shells etc/host.conf etc/securetty I tried with a few others as well, it seems like only the first entry get a md5sum.
On Sat, Aug 25, 2007 at 03:59:36PM +0200, Xavier wrote:
After installing filesystem, here is what I get in the database : %BACKUP% etc/fstab 0d3cfed6eacf21072e468bb7fe80cf07 etc/crypttab etc/group etc/hosts etc/ld.so.conf etc/passwd etc/shadow etc/gshadow etc/resolv.conf etc/motd etc/nsswitch.conf etc/shells etc/host.conf etc/securetty
I tried with a few others as well, it seems like only the first entry get a md5sum.
Well, looks like a little typo was made when copy/pasting code on that last big refactoring :) I think this is probably totally harmless, it should just cause a few extra .pacnew files to be extracted.
From cf0aa0ca29db4de388a8a5c1d76cc0604547d47e Mon Sep 17 00:00:00 2001 From: Chantry Xavier <shiningxc@gmail.com> Date: Sat, 25 Aug 2007 16:26:41 +0200 Subject: [PATCH] libalpm/add.c : fix backup handling.
During last refactoring, a "continue" somehow became a "return(0)" :) Signed-off-by: Chantry Xavier <shiningxc@gmail.com> --- lib/libalpm/add.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/lib/libalpm/add.c b/lib/libalpm/add.c index 398839a..d361873 100644 --- a/lib/libalpm/add.c +++ b/lib/libalpm/add.c @@ -519,7 +519,7 @@ static int extract_single_file(struct archive *archive, backups = alpm_list_next(backups)) { char *oldbackup = alpm_list_getdata(backups); if(!oldbackup || strcmp(oldbackup, entryname) != 0) { - return(0); + continue; } char *backup = NULL; /* length is tab char, null byte and MD5 (32 char) */ -- 1.5.2.5
participants (4)
-
Andrew Fyfe
-
Dan McGee
-
Roman Kyrylych
-
Xavier