Re: [gentoo-dev] ACCEPT_LICENSE revisited

[Marien Zwart]

On Tue, Nov 21, 2006 at 04:37:39PM -0500, Chris Gianelloni wrote:
> Well, we specifically didn't allow a "*" setting because of this.

Ah, I missed that. Thanks.

> Perhaps we should make it simple and specify that no interactive license
> should belong to a group?  That would mean that since we don't include
> it in a group, it won't be part of the "wildcard" NON-INTERACTIVE (or
> whatever it's called) which would make the behavior the same as we
> currently have with check_license, since I think adding group support to
> check_license would be pointless when we're trying to replace it.

I think that would be a good idea. Alternatively portage could export
ACCEPT_LICENSES with the groups expanded. I think that would be
slightly less confusing, although I agree it will probably not come up
in practice (since it is not that likely that licenses used with
check_license will be used in a group). But relying on that not
happening would be a bit icky.

Am I correct in assuming that check_license will be phased out
"eventually" (at some undefined time when everyone runs a portage
supporting ACCEPT_LICENSE)? Perhaps it would be a good idea to include
some information about how this new portage feature interacts with
ACCEPT_LICENSE in the glep (I am assuming more people than just me
were not aware check_license checked the ACCEPT_LICENSE env var)? That
is, explain licenses included in ACCEPT_LICENSE cause check_license to
be "silent", and explain if new ebuilds should be using it or not?

-- 
Marien.