[pacman-dev] Interesting side-effect of IgnorePkg

[Nagy Gabor]

>> Did you try with pacman-git? This should be fixed, see FS#12059.
>
> Haha, of course not. I probably should have dug a bit first, whoops. I
> just built pacman-git on that box and it seems to have done what I
> expected now, so thanks for letting me know I don't even know what
> we've done in 3.3.

It is good that you bring this up, because that part of pacman is  
still ugly. And now I realized that the bug is not completely fixed.  
Why? Pacman calls alpm_grp_get_pkgs(), then the _names_ of the  
packages in the groups are added to the target list (-S pkg1 pkg2...).  
This is not perfect, because it is not guaranteed that we indeed add  
the expected package:
1. First of all, we should use repo/pkg1 to be sure that pacman won't  
add an other package named pkg1.
2. Alpm handles provisions. If we have a gcc provider in sync repos,  
pacman will select that package and will not return with  
PM_ERR_PKG_IGNORED. We can change this behavior but then probably we  
have to reimplement _alpm_resolvedep() in sync_addtarget. (Maybe that  
would be nicer, because we have an ugly prompt param already.)

Personally I vote for cleaning up this part:
Either introduce trans_addtarget_pmpkg(pmpkg_t *pkg) function to  
easily add group packages or move "-S group" handling to the back-end.  
But then we should also do that for -R and probably we want to add  
flags to enable/disable provision/group handling with trans_addtarget.  
Too much work :-( Alternatively, we can disable provision handling in  
sync_addtarget, then it should be also moved to front-end (we need  
more public functions) or introduce trans_addtarget_provision().

> On a side note, I *really* need to make time for 3.3 release planning
> and the actual release. Hopefully everyone on the list will be ready
> for the barrage of activity needed to get this thing out the door, so
> start formulating any thoughts for last minute things that we need to
> fix or do.

See above. ;-) I have one more TODO: resolvedeps.

Bye

------------------------------------------------------
SZTE Egyetemi Konyvtar - http://www.bibl.u-szeged.hu
This message was sent using IMP: http://horde.org/imp/