Andrew Gaffney wrote:
Flag shouldn't be forced, period imo.
I could not agree with this more. I've been watching this whole thread
wondering when certain people were gonna see how dumb this idea actually
is (no offense...I've backed my fair share of dumb ideas). It's a crappy
alternative for Alec's IUSE defaults.
Can you elaborate on how default use flags relate to forcing/masking flags? I'm
just too dumb to see the relation, other than they're both about use flags...
That being said, I would have liked this feature already 1.5 years ago. If we
had had use.force at that time, we didn't have to introduce a
has_multilib_profile function and mask the multilib use flag which lead to much
confusion amongst the users, because up to 2004.3 we told them that they would
be unable to get anything 32bit running without that flag and then suddenly they
all saw (-multilib) in their emerge -pv output. We could have simply continued
to use that flag, forcing it on multilib-enabled profiles and masking it on the
no-multilib ones.
Other examples are probably 'selinux' on selinux-profiles, 'pic' on hardened
(not sure about that one), and ip28 on ip28 profiles (as spb pointed out on IRC).
You may argue that there are not many USE flags, but we have a prove that the
concept would be useful, and the work's basically already been done, so let's
use it.
I know there will be cases where someone will want to force a flag on
for one (or more) package but there will be other packages that forcing
the same flag on is undesireable. Unless use.force can be done
per-package, it will always be a very crappy alternative to IUSE
defaults. Even then, it should still *act* like IUSE defaults (stuck
somewhere in the USE stacking order and "easily" overridden).
I only see this useful for local USE flags which you want to force, like e.g.
ip28. That being said, Zac's patch already handles that case, AFAIK.
--
Kind Regards,
Simon Stelling
Gentoo/AMD64 Developer
--
gentoo-portage-dev@gentoo.org mailing list