Re: [gentoo-dev] Arches marking ebuilds stable before maintainer

[Donnie Berkholz]

Ciaran McCreesh said:
> Besides, if a package version is really that buggy, why isn't it in
> package.mask? The ~ keyword shouldn't be used for dodgy experimental
> releases, it should be used for packages which are candidates to become
> stable after a reasonable (package and arch dependent) period. If an
> ebuild is in the tree and not-*/.mask'ed, we consider it a fair target
> for going stable.

Consider that large numbers of people actually need to test things before
it's known whether they're stable candidates. This does not happen in
package.mask. I consider ~$MYARCH packages fair game for continuing
changes in the voyage toward stability, after bugs are revealed in their
original states.

I will admit, however, that I have a bit of a double standard in when I
bump revisions. The larger the package, the larger the magic number needs
to be for a revision bump. I use a simple, subjective quasi-formula for
this:
( people affected ) * ( severity of change ) / ( average upgrade time )

Example: a minor bug affects a non-default configuration for big-endian
architectures on xorg-x11. Do I bump for the fix? Doubtful.
Example: same situation on colordiff, a package that takes less than a
minute to upgrade. Do I bump? Much more likely.

I realize that colordiff in particular isn't likely to hit that, but it's
just a random situation.

Thanks,
Donnie



--
gentoo-dev@g.o mailing list