- Pacman-dev - lists.archlinux.org

[pacman-dev] What I'm working on- feedback welcome
by Dan McGee 29 Jun '07

29 Jun '07

I've got quite a few branches going on my GIT tree at the moment, and I'd appreciate any feedback from some of you on the list. They are all accessible here: http://code.toofishes.net/gitweb.cgi. * alpm.c_cleanup- move all of the alpm_pkg functions to package.c, alpm_db functions to db.c, etc. Basically put things where they logically belong leaving only the library initialization functions in alpm.c. This is ready to merge; I'm waiting because it could potentially break a lot for anyone else that has something that touched this file. * alpm_list_speed- this was just a test branch to see how much we were losing in efficiency by not having O(1) access to both ends of a list. Answer: a lot. pacman -Qo operations are a lot faster on this branch. However, this isn't meant for merge, it was more of a quick hack-job. * db_test- Add a pacman -Qt operation which tests the integrity of your database; ported over from the Frugalware pacman-g2. * frontend_config- move the pacman.conf parsing to the frontend where it belongs. Move a few pacman-only settings out of the library. In general, try to clean up where options and configs are stored. * gettext- gettext the remaining scripts in the scripts/ directory. I've had these patches sitting around for a while but never got them applied. Ready to be merged. * working- remove some hardcoded paths from the pacman binary, making everything configurable at runtime instead of compile time through pacman.conf. This means the only path now hardcoded into the pacman binary is that to pacman.conf, and no paths are compiled into libalpm. As I said, feedback on any of these is welcome and appreciated. -Dan

2 1

[pacman-dev] Pacman depending on package naming scheme
by Dan McGee 29 Jun '07

29 Jun '07

If anyone on the list wants to look this commit/branch of mine over and review it, I'd appreciate it: http://code.toofishes.net/gitweb.cgi?p=pacman.git;a=shortlog;h=pkgname_check I know Andrew just had a commit dealing with a very similar thing in the xdelta creation in makepkg, and he removed the dependency on a package having a certain name. This is the same sort of issue, and I think it would be wise for us to fix it everywhere we can. Not only does this free us up from the CARCH issue, note the amount of code we can remove because of it. (For a side note- I challenge anyone to start submitting patches to the list that result in a net removal rather than addition of code. They are a lot more helpful than you would think.) Does anyone have concerns with this, and does anyone have some great ideas how to (re)implement partial cache cleaning functionality correctly? The old way wasn't phenomenal, and I just got rid of it in this first commit here. There were also many shortcomings in it as you can see in the comments throughout that section. -Dan

2 1

[pacman-dev] My git repo and development snapshots
by Andrew Fyfe 29 Jun '07

29 Jun '07

I've finally got round to getting my self a vps server @ slicehost.com, so I've now got all my git stuff running on the new server. I'll keep my repo at repo.or.cz up to date. Also for those that want to get the latest source without git I've now got git generating snapshots of the master branch. They're automatically updated when new updates are posted.* git: git://neptune-one.homeip.net/pacman gitweb: http://neptune-one.homeip.net/cgi-bin/gitweb.cgi?p=pacman snapshots: http://neptune-one.homeip.net/~andrew/pacman/ Andrew * More specifically when I push updates, there doesn't seem to be a hook for git pull :(

2 1

[pacman-dev] [GIT] pacman branch, master now at v3.0.0-254-g78bac81
by Andrew Fyfe 29 Jun '07

29 Jun '07

Hello, This is an automated email from the git hooks/update script, it was generated because a ref change was pushed to the repository. Updating branch, master, via 78bac81a58355a6d42760e1974005dbcdd8e8dcf (commit) from 7daa6708d2afc710bebbb2dc30f4371f9d67120c (commit) - Log ----------------------------------------------------------------- commit 78bac81a58355a6d42760e1974005dbcdd8e8dcf Author: Andrew Fyfe <andrew(a)neptune-one.net> Date: Thu Jun 28 10:48:32 2007 +0100 scripts/makepkg.in: Don't include .PKGINFO etc in .FILELIST And move .PKGINFO and .FILELIST to be the first to files included in the package. Signed-off-by: Andrew Fyfe <andrew(a)neptune-one.net> ----------------------------------------------------------------------- Diffstat: scripts/makepkg.in | 11 +++++++---- 1 files changed, 7 insertions(+), 4 deletions(-) hooks/update --- Git Source Code Management System hooks/update refs/heads/master \ 7daa6708d2afc710bebbb2dc30f4371f9d67120c \ 78bac81a58355a6d42760e1974005dbcdd8e8dcf

3 3

[pacman-dev] libarchive warning output
by Xavier 28 Jun '07

28 Jun '07

There is one more change with libarchive 2.2.3 vs 1.3.1 When installing a package as user, archive_read_extract now returns ARCHIVE_WARN instead of ARCHIVE_OK , because it cant set uid to 0, and also, it can't set suid for the files that have it. Currently, pacman considers it as errors, instead of warnings. I don't think it should, but I'm not sure how to handle the warning output either. Also, strerror(errno) displayed either "File exists" or "No such file or directory", I've no idea why... And archive_error_string(archive) still displays that part, but also adds more helpful message like : "Can't set UID=0: File exists" "Can't set UID=0: No such file or directoryCan't make file SUID" Since with old libarchive, these warnings weren't returned, and so not displayed, I adpated pacman to keep the same behavior for now. But if these warnings are useful, they can easily be displayed. >From 2657a2bfb7847c8ac4dd5875979400d0d3fd89f6 Mon Sep 17 00:00:00 2001 From: Chantry Xavier <shiningxc(a)gmail.com> Date: Thu, 28 Jun 2007 19:28:34 +0200 Subject: [PATCH] libalpm/add.c : ignore libarchive warning. With libarchive 2.2.3 (previously 1.3.1), archive_read_extract now returns ARCHIVE_WARN when a package is extracted as user, because for example, UID=0 or SUID bit can't be set. This patch makes pacman not treating these warnings as errors anymore, but simply ignoring them. Signed-off-by: Chantry Xavier <shiningxc(a)gmail.com> --- lib/libalpm/add.c | 17 +++++++++++------ 1 files changed, 11 insertions(+), 6 deletions(-) diff --git a/lib/libalpm/add.c b/lib/libalpm/add.c index b6a54b7..1ebf455 100644 --- a/lib/libalpm/add.c +++ b/lib/libalpm/add.c @@ -544,11 +544,12 @@ int _alpm_add_commit(pmtrans_t *trans, pmdb_t *db) archive_entry_set_pathname(entry, tempfile); - if(archive_read_extract(archive, entry, archive_flags) != ARCHIVE_OK) { + int ret = archive_read_extract(archive, entry, archive_flags); + if(ret != ARCHIVE_OK && ret != ARCHIVE_WARN) { _alpm_log(PM_LOG_ERROR, _("could not extract %s (%s)"), - entryname, strerror(errno)); + entryname, archive_error_string(archive)); alpm_logaction(_("could not extract %s (%s)"), - entryname, strerror(errno)); + entryname, archive_error_string(archive)); errors++; unlink(tempfile); FREE(hash_orig); @@ -699,10 +700,14 @@ int _alpm_add_commit(pmtrans_t *trans, pmdb_t *db) archive_entry_set_pathname(entry, filename); - if(archive_read_extract(archive, entry, archive_flags) != ARCHIVE_OK) { - _alpm_log(PM_LOG_ERROR, _("could not extract %s (%s)"), filename, strerror(errno)); - alpm_logaction(_("error: could not extract %s (%s)"), filename, strerror(errno)); + int ret = archive_read_extract(archive, entry, archive_flags); + if(ret != ARCHIVE_OK && ret != ARCHIVE_WARN) { + _alpm_log(PM_LOG_ERROR, _("could not extract %s (%s)"), + entryname, archive_error_string(archive)); + alpm_logaction(_("could not extract %s (%s)"), + entryname, archive_error_string(archive)); errors++; + continue; } /* calculate an hash if this is in newpkg's backup */ -- 1.5.2.2

1 0

[pacman-dev] Pulled changes and other things
by Dan McGee 28 Jun '07

28 Jun '07

I pulled in Andrew's bsdtar branch today, and Xavier's ready_to_pull branch, so those changes have all been pushed to the al.org tree. Thanks guys! Other things: 1. For you guys that I am pulling GIT repos from, can you do me a favor? I added some more descriptive tags to the GIT repo (v3.0.0) instead of the older uglier CVS tags (R_3_0_0). If you wouldn't mind deleting those old tags in your pushed repo, I would appreciate it so I don't keep pulling them in. By default GIT does not push tags, but it does pull them. I got rid of them on my public repo and the al.org tree; if you do the same that would be awesome. 2. What is on people's minds for a 3.1 release? I don't think its too early to start at least getting some thoughts on paper, and see what kind of timeframe we can make for ourselves. I know at least the following things are true- I've made a lot of changes recently and they need some refinement, to say the least. In addition, we seem to have a lot of outstanding libarchive issues that we should take a look at and ensure are fixed. If we need to extend the pactest framework to be able to test this better, then we should do that. Thanks to everyone that has been contributing to pacman, we definitely appreciate it, and you can see the wide variety of authors that are now in the GIT history. Hopefully some of you will venture into the big bad libalpm after dipping your toes in makepkg changes. It is what I did. :) -Dan

3 2

[pacman-dev] Color patch again
by Vojtěch Gondžala 28 Jun '07

28 Jun '07

Hi all, I rewrite the color patch from 3.0.x version. Colors are reconfigurable in pacman.conf in section [colors], if isn't anything explicity named then is using default color. Option --nocolor disable colorized output. yours opinion? :) --vogo

3 3

[pacman-dev] [GIT] pacman branch, master now at v3.0.0-253-g7daa670
by Dan McGee 28 Jun '07

28 Jun '07

Hello, This is an automated email from the git hooks/update script, it was generated because a ref change was pushed to the repository. Updating branch, master, via 7daa6708d2afc710bebbb2dc30f4371f9d67120c (commit) via 7bdb904af5b664c1d6e47977f21f0e6bc3f8e3b0 (commit) via ad691001e20272b794d2ed574b556f520e3555c0 (commit) from 1e56e41d638fcdddc291d9a8ea183b40f4802aae (commit) - Log ----------------------------------------------------------------- commit 7daa6708d2afc710bebbb2dc30f4371f9d67120c Author: Dan McGee <dan(a)archlinux.org> Date: Wed Jun 27 23:25:04 2007 -0400 Remove lockfile configuration from frontend, make it job of libalpm I previously introduced some patches to make just about every path in pacman/libalpm configurable; doing this with the lockfile seemed a bit too far and we really should just place the lockfile where it belongs- with the DB that needs locking. More details in this thread: http://archlinux.org/pipermail/pacman-dev/2007-June/008499.html Signed-off-by: Dan McGee <dan(a)archlinux.org> commit 7bdb904af5b664c1d6e47977f21f0e6bc3f8e3b0 Author: Dan McGee <dan(a)archlinux.org> Date: Wed Jun 27 22:58:21 2007 -0400 pacman.c: make parseconfig a bit more robust Don't let parseconfig overwrite settings that parseargs already made. Signed-off-by: Dan McGee <dan(a)archlinux.org> commit ad691001e20272b794d2ed574b556f520e3555c0 Author: Dan McGee <dan(a)archlinux.org> Date: Wed Jun 27 22:12:59 2007 -0400 Remove scriptlet START and DONE commands that we don't use The scriptlet calling had some unneeded complexity for the time being which we aren't using here. Let's get rid of it until we find a good way to implement it correctly. Signed-off-by: Dan McGee <dan(a)archlinux.org> ----------------------------------------------------------------------- Diffstat: configure.ac | 1 - etc/pacman.conf.in | 1 - lib/libalpm/Makefile.am | 5 ----- lib/libalpm/alpm.h | 5 +---- lib/libalpm/handle.c | 17 +++++++---------- lib/libalpm/trans.c | 33 +-------------------------------- lib/libalpm/util.c | 9 +++------ lib/libalpm/util.h | 4 ---- pactest/pmtest.py | 5 ++--- pactest/util.py | 2 +- scripts/pacman-optimize.in | 2 +- src/pacman/Makefile.am | 1 - src/pacman/callback.c | 14 -------------- src/pacman/pacman.c | 35 +++++++++++++++++------------------ 14 files changed, 33 insertions(+), 101 deletions(-) hooks/update --- Git Source Code Management System hooks/update refs/heads/master \ 1e56e41d638fcdddc291d9a8ea183b40f4802aae \ 7daa6708d2afc710bebbb2dc30f4371f9d67120c

1 0

[pacman-dev] [GIT] pacman branch, master now at v3.0.0-250-g1e56e41
by Dan McGee 28 Jun '07

28 Jun '07

Hello, This is an automated email from the git hooks/update script, it was generated because a ref change was pushed to the repository. Updating branch, master, via 1e56e41d638fcdddc291d9a8ea183b40f4802aae (commit) via 269022ccae921ae7f7caa1aca4a7d40cb0773985 (commit) via 24c41f5190974e68444466d0d8563dbe9ab06768 (commit) via 5a3b59583740c033a5e27b61c3c38bebfd580acc (commit) via 77bbe581973d41d57edb96488fa2cf73fddc1641 (commit) via 3a27fbaae40869d513cf117609d3a56c07863cae (commit) from d821d75e255207bf84f58532e2ff65a8a2a79d42 (commit) - Log ----------------------------------------------------------------- commit 1e56e41d638fcdddc291d9a8ea183b40f4802aae Author: Andrew Fyfe <andrew(a)neptune-one.net> Date: Wed Jun 27 12:04:33 2007 +0100 scripts/makepkg.in: Use bsdtar to find packages for xdelta. When looking for old packages to create a delta from used bsdtar to read the packages .PKGINFO to check it's name and arch instead of depending on the filename of th package. Signed-off-by: Andrew Fyfe <andrew(a)neptune-one.net> commit 269022ccae921ae7f7caa1aca4a7d40cb0773985 Author: Chantry Xavier <chantry.xavier(a)gmail.com> Date: Fri Jun 22 22:09:55 2007 +0100 scripts/makepkg.in: Fix creation of .FILELIST Signed-off-by: Andrew Fyfe <andrew(a)neptune-one.net> commit 24c41f5190974e68444466d0d8563dbe9ab06768 Author: Baptiste Daroussin <baptiste.daroussin(a)gmail.com> Date: Fri Jun 22 19:23:28 2007 +0100 FS7485: Replace tar/cpio/unzip with bsdtar for extracting/creating packages. Signed-off-by: Andrew Fyfe <andrew(a)neptune-one.net> commit 5a3b59583740c033a5e27b61c3c38bebfd580acc Author: Dan McGee <dan(a)archlinux.org> Date: Wed Jun 27 19:27:26 2007 -0400 Fix wrong permissions on pacnew extraction First reported here: http://bbs.archlinux.org/viewtopic.php?pid=261861 Newly created files were done with the standard umask, so those that are extracted seperately and copied to a .pacnew extension will have the wrong permissions. This should hopefully fix this. Signed-off-by: Dan McGee <dan(a)archlinux.org> commit 77bbe581973d41d57edb96488fa2cf73fddc1641 Author: Dan McGee <dan(a)archlinux.org> Date: Wed Jun 27 16:33:27 2007 -0400 Remove TODO items that have been taken care of. Signed-off-by: Dan McGee <dan(a)archlinux.org> commit 3a27fbaae40869d513cf117609d3a56c07863cae Author: Chantry Xavier <shiningxc(a)gmail.com> Date: Thu Jun 28 01:20:30 2007 +0200 libalpm/remove.c : file variable used before being initialized. Signed-off-by: Chantry Xavier <shiningxc(a)gmail.com> ----------------------------------------------------------------------- Diffstat: TODO.aaron | 16 ---------------- TODO.dan | 17 ----------------- lib/libalpm/remove.c | 3 ++- lib/libalpm/util.c | 15 +++++++++++++++ scripts/makepkg.in | 44 +++++++++++++++++--------------------------- 5 files changed, 34 insertions(+), 61 deletions(-) hooks/update --- Git Source Code Management System hooks/update refs/heads/master \ d821d75e255207bf84f58532e2ff65a8a2a79d42 \ 1e56e41d638fcdddc291d9a8ea183b40f4802aae

1 0

[pacman-dev] makepkg creates bad .FILELIST
by Andrew Fyfe 27 Jun '07

27 Jun '07

This patch[1] breaks the creation of .FILELIST, directories don't end with / so pacman treats them as files and does conflict checks and bails out because /usr (or any other dir) has already been installed by another package. Posted a patch[2] to my ready_to_pull branch. [1] http://projects.archlinux.org/git/?p=pacman.git;a=commit;h=430a19c604e474e9… [2] http://repo.or.cz/w/pacman.git?a=shortlog;h=ready_to_pull Andrew

4 11