[arch-dev-public] Strange behaviour of pacman

[Aaron Griffin]

On Mon, Oct 19, 2009 at 2:21 PM, Daniel Isenmann <daniel.isenmann at gmx.de> wrote:
> On Mon, 19 Oct 2009 14:08:33 -0500
> Aaron Griffin <aaronmgriffin at gmail.com> wrote:
>
>> On Mon, Oct 19, 2009 at 2:03 PM, Daniel Isenmann
>> <daniel.isenmann at gmx.de> wrote:
>> > On Mon, 19 Oct 2009 14:45:20 -0400
>> > Eric Bélanger <snowmaniscool at gmail.com> wrote:
>> >
>> >> On Mon, Oct 19, 2009 at 2:06 PM, Travis Willard
>> >> <twillard2 at gmail.com> wrote:
>> >> >> As I can see now, these are .pyo files. Are they generate at
>> >> >> runtime or something like that? They are not in the package.
>> >> >>
>> >> >
>> >> > .pyo files are, I believe, "optimized" python files generated
>> >> > during runtime.
>> >> >
>> >>
>> >> I beleiie so too.  I think there was a thread about how to deal
>> >> with these files.  I think the info is in a wiki article about
>> >> python packaging guidelines. The other remaining file is wicd.log
>> >> wich is generated at runtime too.
>> >
>> > I have nothing found about those files. The article about python
>> > package guidelines is very short. Nothing special about it.
>> >
>> > The log file is acceptable, but the pyo files are annyoing.
>>
>> I imagine that this only happens with apps run as root (or have write
>> permissions to their install dir).
>>
>> I think the best thing, for the time being, is to do this in a
>> pre_remove (so you have access to pacman -Ql at that time) and do
>> something like:
>>
>> PKGNAME=wicd
>> pre_remove () {
>>     for pyo in $(pacman -Qql $PKGNAME  | grep \.py$ | sed
>> 's|.py$|.pyo|g'); do if [ -f "$pyo" ]; then
>>             rm "$pyo"
>>         fi
>>     done
>> }
>
> Ok, I will do it this way, but shouldn't we have a better solution for
> this for the future?

Well, the only sane way to do it would be to make sure pacman tracks
the .pyo files by generating them as part of the package creation
process, but I don't even know if that's possible