Re: Proposed patch management/build solution

[Sam Hartman <hartmans@debian.org>]

>>>>> "Joel" == Joel Baker <lucifer@lightbearer.com> writes:


    Joel> I'll worry about this when it doesn't require a chroot
    Joel> tarball just to get things up and running. Until then... the
    Joel> only folks who're likely to be even remotely interested in
    Joel> debugging are the diehards, who can find all they need in
    Joel> the list archives.

The current state of disorganization is such that I'd rather either
organize or just not bother contributing than to hack together a
debugging environment.  I don't consider myself to be completely
unfamiliar with OS installation/distribution development; I am
certainly capable of digging through list archives and grabbing things
from three or four different places, but really I'd rather not.



    Joel> I *strongly* advise you to look at the existing auto-patch
    Joel> system that is in place.

Explain auto-patch system, preferably with a pointer.


    >> Then we can set up this CVS repository somewhere (presumably
    >> initially sourceforge) and allow people to start committing
    >> their patches.  I think this organization will really help me
    >> be able to debug things and make forward progress; will it help
    >> others enough to be used?

    Joel> How will you handle detection of patches appearing upstream?

I do not understand what you mean here?

    Joel> Who has the authority to remove patches? 

This question only becomes a problem if there is some action that  we is taken that cannot be validated by consensus.  The likelihood that the project will be producing enough work that it is viable, that no clear leaders will emerge, and that  such a action is taken is fairly low.


    Joel> Will there be an
    Joel> autobuilder working from this tree, and, if so, where will
    Joel> the packages appear?  --


I don't anticipate a Debian style autobuilder at least immediately.
EVentually yes, but it may well be that we no longer need this system
by the time we are stable enough for a formal autobuilder.  I'm
certainly willing to build packages from this tree and make them
appear in MIT AFS space, which is web accessible.