[gentoo-dev] Re: don't rely on dynamic deps

[Martin Vaeth]

Ulrich Mueller <ulm@g.o> wrote:
> The new ebuild is cat/foo-1-r0.1 then, and PF changes even
> if the minor revision is ignored (namely, from foo-1 to foo-1-r0).

PF has to be filled correctly, of course:
The versions foo-1 and foo-1-r0 are identical according to PMS
and should thus lead to the same $PF.

If this is currently not happening in portage, this is a bug
which should be fixed, independently of whether minor revisions
are introduced or not.

Note that PR=0 in both cases.

BTW: Even if PF would have the wrong value, this would not do
any harm for minor revisions: If the minor revision has just
changed, nothing is merged to the filesystem and /var/db/*/*/CONTENTS
is unchanged, either. So nothing breaks in this case.
If the user re-emerges the minor revision (and if PF would have
the wrong value), then the minor revision would be visible to the
user in the sense that the name of the Doc-Directory is influenced
(and after a further minor revision bump, this revision number
would be outdated). An arguable cosmetical issue,
but nothing breaks, either.

>> Actually, I still think that implementing dynamic deps correctly
>> would be better, but minor revisions do not exclude this
>> and are probably simpler to implement.
>
> The PM could just update the VDB whenever it detects that the metadata
> of the ebuild has changed (relative to the VDB). You can get that
> without introducing the additional complication of minor revisions.

...but by introducing all the additional complications Ian
has mentioned. More precisely: What happens if new dependencies
are introduced which are not satisfied?
One has to face it: Portage must not just silently "update" the
database and thus silently produce a /var/db which does not
satisfy its own dependencies...