[pacman-dev] Fix reason handling before 3.1 (was: Final steps before 3.1 release)

Nagy Gabor ngaba at bibl.u-szeged.hu
Mon Dec 3 12:08:08 EST 2007

Idézés Dan McGee <dpmcgee at gmail.com>:

> On Dec 3, 2007 4:47 AM, Nagy Gabor <ngaba at bibl.u-szeged.hu> wrote:
> > Well, the changelog between 3.0.x and 3.1 will be huge and impressive;-)
> > There are some issues which annoy me a bit:
> > -sync044.py (but this is only one thing: see
> > http://www.archlinux.org/pipermail/pacman-dev/2007-September/009271.html
> for
> > details) <- %REASON% is one of my favourite libalpm features, and now this
> is
> > quite buggy :-(
> > -FS#8350
> >
> > OK, I don't want to delay this release (imho we should release pacman more
> > frequently <- but then we should do a message freeze between two major
> versions
> > to help to our translators), but I think we should release 3.1.1 soon to
> fix
> > these. Or to be precise, these are libalpm (libdownload?) issues, but in AL
> we
> > have no separate libalpm package.
> I think we can get this into 3.1 without a problem, IF we can exactly
> lay out the problem. From your other email, it looks like we should
> have a sequence of patches that do the following:
> 1. Add an --asexplicit option
> 2. First, determine a REASON policy for things such as installs,
> upgrades, and syncs including pulling in deps. Second, make a patch
> that implements this (and has comments in the code and commit msg!)
> 3. Implement HoldPkg reason (value of 2)? If they aren't already,
> these should become constants/enums in the code since we now have more
> than two possibilities.
> So, in short Nagy- pick what you want to do above and help us get this
> done. Also, let me know what Aaron or I need to do to help you.
> -Dan

3. is not important (I misinterpreted Holdpkg)

Well, I overlooked something: "it owerwrites the old reason iff the new reason
is depend" <- now this is not true imho (pactests?).

%REASON% communication between upgrade and sync transactions is chaotic (via
pkgcache), but looks OK:
-sync.c presets reason: dependency iff this is a dependency (bit buggy, sync044.py)
-add.c keep this result if no old version was found (package born), and copy the
old reason if old version was found

1. is easy to implement, but we are in message freeze (or not?)

2. this needs sync.c (pmsyncpkg_t) rework, but currently sync.c is so
hard-to-understand that I don't want to do this.


SZTE Egyetemi Könyvtár - http://www.bibl.u-szeged.hu
This mail sent through IMP: http://horde.org/imp/

More information about the pacman-dev mailing list