On Fri, 04 Mar 2005 15:01:52 +0100, Bastian Balthazar Bux wrote:

> IUSE are not defined only in ebuils but also in eclass,
> and a new use flag not implyes a version bump.
> And the other half of a point, the profile change apply to quite every 
> relase.

But you wouldn't change a profile without knowing it. It is a manual
action that can be followed with emerge --newuse. It is not something that
needs to be addressed from a daily cron job.


> [quote]
> #  Try to not bump ebuilds continuously unless there really is a benefit
> or a security fix which is important. Unnecessary examples of bumping 
> include:
> 
>      * You change minor spelling errors in script file comments, script 
> file indentation or something similar.
>      * You patch a non-kernel ebuild to support a new kernel version (or
>      
> a new version of a library), allowing more users to install your ebuild,
> 
> but not changing anything for existing users of the current revision.
> 
> As a general rule, fixes with non-trivial changes to any of the 
> installed files of any ebuild warrant a revision bump. Put differently: 
> If your fix changes the behaviour for existing users, you bump so that 
> they know they can upgrade.
> [/quote]

Changing USE flags generally changes the behaviour, so I would expect it
to result in a version bump. If it doesn't change the behaviour for
existing users, there is no need to recompile, which is what this
recommendation is all about.

But this whole discussion is moot, as esync doesn't use --newuse and the
alias I suggested was an eix replacement for esync.


-- 
Neil Bothwick

"Bother," said Pooh, as the vice squad took his GIFS

Attachment: pgpg11KuAGQUU.pgp
Description: PGP signature

Reply via email to