Re: [pacman-dev] [PATCH] -Qu rework

2008/11/1 Nagy Gabor :

Sat, 1 Nov 2008 08:19:51 -0500 -n "Dan McGee" írta:

...

...

...

Hmm, this isn't quite what I expected either. Can we clean up this garbage in any way?

$ ./src/pacman/pacman -Qu warning: alsa-lib: local (1.0.17a-2) is newer than extra (1.0.17a-1) warning: gvim: local (7.2.25-1) is newer than extra (7.1.330-1) warning: namcap: local (2.1-2) is newer than extra (2.1-1) warning: openoffice-base: local (3.0.0-3) is newer than extra (3.0.0-2) warning: pacman-git: local (20081031-1) is newer than pacman-git-64 (20081028-1) warning: pycairo: local (1.6.4-2) is newer than extra (1.6.4-1) warning: pygobject: local (2.15.4-2) is newer than extra (2.15.4-1) warning: pygtk: local (2.13.0-2) is newer than extra (2.13.0-1) warning: python: local (2.6-2) is newer than extra (2.5.2-5) warning: python-numeric: local (24.2-3) is newer than extra (24.2-2) warning: sonata: local (1.5.3-2) is newer than extra (1.5.3-1) warning: vi: local (7.2.25-1) is newer than core (7.1.330-1) warning: vim: local (7.2.25-1) is newer than extra (7.1.330-1) warning: vte: local (0.17.4-2) is newer than extra (0.17.4-1)

-Dan

1. Since I am lazy, I first try with the easiest solution. What about completely removing these messages? Are these useful with -Su at all? (The old -Qu printed these messages as well, bacause it was a simulation of -Su). I personally find these messages helpful in the -Su case- when crazy

On Sat, Nov 1, 2008 at 8:16 AM, Nagy Gabor wrote: things like the above come up, you know something is up with your mirrors and whatnot.

...

2. We could disable all warnings like with -Sp. Never was too fond of this either. Seems hackish as is.

...

3. Most complicated: Give a new parameter to sync_newversion. This could fix the possible duplicated "pacman is newer than extra" message on -Su. This seems really ugly.

Wow, I'm a negative Nancy here. What if we proceed with 1 (so remove these things from alpm_sync_newversion), but somehow in sysupgrade we still can have this relevant info? I don't know if that is even possible.

Without 3. this seems impossible. Then we should reimplement sync_newversion inside sync_sysupgrade (and may rename sync_newversion to ~pkg_outdated) or add a new out param (cmp) to it... Btw, sync_newversion is quite simple function (until we don't merge replacement stuff into it. I referred to FS#11737 here.)

My vote is 2., even if it is hackish. In my mind "warnings" are just "additional info" messages, they never indicate important things (those are errors).

This just got me tonight. Should we just go with two, or can we buffer things to order them right or something? I'm not sure. The other option would be to use stdout and stderr in a slightly better fashion. -Dan