Re: [pacman-dev] [arch-general] [arch-dev-public] Versioned dependency fun...
On Tue, Mar 2, 2010 at 4:40 PM, Xavier Chantry <chantry.xavier@gmail.com> wrote:
On Tue, Mar 2, 2010 at 7:32 PM, Pierre Schmitz <pierre@archlinux.de> wrote:
Am Dienstag, 2. März 2010 14:06:57 schrieb Dan McGee:
and here we have another reason to punish anybody using versioned dependencies. ;-) (of course upstream needs to be hit, too)
Blaming this on versioned deps is punishing the messenger; this package wouldn't upgrade without a force option either because the same version comparison is used everywhere...
You did notice the smiley? I know this wasn't fair in this case and not too serious. But anyway, I would prefer add the force flag and add the (stupid) upstream version instead of making your own up.
The force flag is a big ugly crap. I find sanitizing upstream version to be much.. saner. Or figure out a way to have both a sane version for pacman and the insane upstream version.
I was actually thinking about this again and thinking epoch might not be a bad idea- I don't know if I had any legit objection to it before but I can't think of one anymore. If someone is interested in implementing, I'd be for it. -Dan
On 03/03/10 13:59, Dan McGee wrote:
I was actually thinking about this again and thinking epoch might not be a bad idea- I don't know if I had any legit objection to it before but I can't think of one anymore. If someone is interested in implementing, I'd be for it.
I can not remember any objections either. From memory, the main discussion was how best to implement it but got side-tracked by the need to push a release. Allan
Am 03.03.2010 04:59, schrieb Dan McGee:
The force flag is a big ugly crap. I find sanitizing upstream version to be much.. saner. Or figure out a way to have both a sane version for pacman and the insane upstream version.
I was actually thinking about this again and thinking epoch might not be a bad idea- I don't know if I had any legit objection to it before but I can't think of one anymore. If someone is interested in implementing, I'd be for it.
I was thinking more of having two version numbers. The pkgver and the optional, internal sanitized pkgver. The internal one is used for comparison if it exists, but pacman always displays the pkgver for upstream consistency. Maybe epoch is better, but I wanted to mention this idea.
On Wed, Mar 3, 2010 at 2:35 AM, Thomas Bächler <thomas@archlinux.org> wrote:
Am 03.03.2010 04:59, schrieb Dan McGee:
The force flag is a big ugly crap. I find sanitizing upstream version to be much.. saner. Or figure out a way to have both a sane version for pacman and the insane upstream version.
I was actually thinking about this again and thinking epoch might not be a bad idea- I don't know if I had any legit objection to it before but I can't think of one anymore. If someone is interested in implementing, I'd be for it.
I was thinking more of having two version numbers. The pkgver and the optional, internal sanitized pkgver. The internal one is used for comparison if it exists, but pacman always displays the pkgver for upstream consistency. Maybe epoch is better, but I wanted to mention this idea.
I think Thomas' idea is a good one (using some internal number), and using the epoch for the internal value doesn't sound too bad.
participants (4)
-
Aaron Griffin
-
Allan McRae
-
Dan McGee
-
Thomas Bächler