[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#994388: dpkg currently warning about merged-usr systems

Josh Triplett <josh@joshtriplett.org> writes:

> I think it's appropriate for people to wait on such work until there's
> guidance from the TC ensuring that such a patch will be accepted.
> Otherwise, anyone spending time writing it is spending substantial
> effort that may well be wasted.

I think this is a totally fair thing to be concerned about.  Should such a
patch exist -- with the obvious condition that I think it's quite
reasonable to do several rounds of iteration on making that patch solid,
ensuring there are tests, and so forth -- I think it's obvious that we
should merge it given the previous TC decision.  Of course, I'm not a TC

It's difficult, procedurally, for the TC to do anything about a
theoretical patch that someone could write but hasn't written.
Particularly for dpkg, the details are important.  I can think of some
ways of supporting merged-/usr that I wouldn't support even while
supporting the TC decision.  We have various goals (such as being able to
bootstrap entirely through package installation) that can be met while
supporting merged-/usr but which do require design and care.

If a concrete patch exists, the TC can (and has in the past) authorize an
NMU to apply it.  Obviously, we should try to avoid reaching that level of
social and process confrontation if we can avoid it, but this is clearly
within the TC's constitutional power via a maintainer override, which puts
the discussion on somewhat firmer ground.  But design of that patch is
*not* within the TC's constitutional mandate.

It may be useful to look at how multiarch support in dpkg was handled.
That was quite painful and I really hope we don't end up following that
path exactly, but it provides a concrete example of how Debian's processes
can reach a resolution.

I personally am still hopeful that we could do much better than the
multiarch outcome and find a patch that meets the architectural criteria
of the dpkg maintainer, but I'm fairly certain that we're not going to
make any progress towards that goal without having working code, or at
least a very detailed architecture, to start discussing and analyzing.

Russ Allbery (rra@debian.org)              <https://www.eyrie.org/~eagle/>

Reply to: