Re: [gentoo-dev] Looking for a generic solution to non-USE-conditional circular deps

[Michał Górny]

Dnia 2015-04-11, o godz. 16:00:36
Pacho Ramos <pacho@g.o> napisał(a):

> El sáb, 11-04-2015 a las 14:28 +0200, Michał Górny escribió:
> > Hello,
> >
> > I know of two cases of circular deps that can be solved -- as far as I
> > know -- only by bundling:
> >
> > 1. dev-util/pkgconfig <-> dev-libs/glib:2,
> >
> > 2. dev-util/cmake <-> dev-libs/jsoncpp.
> >
> > The former is solved via having USE=internal-glib on pkgconfig,
> > the latter is yet unsolved [1].
> >
> > I suspect we can expect more of this kind of issues in non-systemd
> > world over time, and I feel like it'd be good to develop a generic
> > policy for handling them. Something that could avoid having users go
> > through this on a case-by-case basis.
> >
> > I think the simplest solution would be to develop a generic USE flag
> > that would only serve the purpose of forcing bundled dependencies for
> > bootstrapping/initial install.
> >
> > We have already:
> >
> > build - !!internal use only!! DO NOT SET THIS FLAG YOURSELF!, used for
> > creating build images and the first half of bootstrapping [make stage1]
> >
> > bootstrap - !!internal use only!! DO NOT SET THIS FLAG YOURSELF!, used
> > during original system bootstrapping [make stage2]
> >
> > However, since both are marked for 'internal use only', I don't think
> > it's a good idea to use them here. So I guess we need a new flag. Does
> > anyone have suggestions how to name it?
> >
> > Any other ideas?
> >
> > [1]:https://bugs.gentoo.org/show_bug.cgi?id=546130
> >
>
> As that USE flag should only be used for being able to install the
> package the first time, maybe it should be treated in a "special" way. I
> mean, it shouldn't be easily changed by users but, instead, switched
> "internally" by the package manager to allow to unbreak the cycle.
>
> Otherwise, we could end up with people having packages built with
> "bundled" libs forever because they enabled that USE at some point and
> later they forgot (or were too lazy) to revert that :|

Yes, that's why I suggest a generic flag rather than another
USE=internal-*.

-- 
Best regards,
Michał Górny