On Sat, 2013-12-28 at 08:05 -0800, W. Trevor King wrote:
> On Sat, Dec 28, 2013 at 06:54:28AM -0500, Mike Frysinger wrote:
> > On Saturday 28 December 2013 01:11:08 Brian Dolbec wrote:
> > > I don't know if catalyst can do that or not and whether it can
> > > detect it to run perl-cleaner.
> > 
> > catalyst certainly could grow code for handling perl-cleaner &
> > friends.
> 
> This sounds like a job for Portage and @preserved-rebuild ;).  It has
> already absorbed revdep-rebuild.  Hopefully absorbing perl-cleaner,
> python-updater, and similar is just a matter of time :p.
> 
> Cheers,
> Trevor
> 

No, it didn't absorb revdep-rebuild.  It just does early (live)
detection.

Actually, with the new python version of revdep-rebuild, there was talk
of it absorbing python-updater and perl-cleaner (also coded in python)
as part of it's normal revdep call.  Then it would be a one stop call.

Actually, I think in this case. the original error would have been
avoided if the user had used the same snapshot for all stages of the
build.  stage1 through livecd-*.  I believe he/she just grabbed a stage3
and a current tree snapshot.

We could however change catalyst to accept a stage3 and perform an
update, perl-cleaner, python-updater, revdep-rebuild,... before
commencing with the build process, but I feel that is a step better done
by hand.  It would also need to wait for catalyst to move to using
emerge/portage through the api.  From there it could get the list of
atoms to be updated and handle any calls on an as needed basis to
perl-cleaner, etc..  That would also require me to complete the
public_api work I was doing for portage.