Re: [Gdbheads] proposed change to GDB maintainership rules

[Daniel Jacobowitz]

On Sat, Jan 31, 2004 at 02:50:57AM -0500, Richard Stallman wrote:
>     Voting is for a last resort conflict resolution, not
>     for everyday approval.  The central point of the proposal
>     is that blanket write maintainers have the authority to
>     approve patches in any part of gdb, even if there is an
>     area maintainer assigned to that part.
> 
> This suggests an smaller change that might address the issue just as
> well: assign more area maintainers to whichever areas are not
> reviewing patches fast enough.  If the GDB Maintainers make sure there
> are enough area maintainers in every area, things should work
> smoothly.  Some global maintainers could be chosen as secondary area
> maintainers, to review patches in those areas when the main area
> maintainers don't.  Some new people could also become area
> maintainers.

I don't think that this is a practical solution to the problem of patch
review as we are currently experiencing it.

The current difficulties are not "we have a set of area maintainers who
are not doing their job".  We've had that problem in the past; it's
been dealt with, usually by Andrew.  The maintainers in question either
shaped up or stepped down.

Instead I believe that the current problem is a network of interlocking
bottlenecks.  No one has time to review patches all the time - for the
current set of GDB maintainers, I'd say that most of the time most of
the people do not have time to review patches.  By having small groups
of maintainers responsible for individual areas, most patches not
proposed by someone who can approve it directly have to sit until
someone has time to review them, and review times are long.

If we double the size of the small groups, it might alleviate the
pressure temporarily, but not solve the problem.  This is especially
true because the pool of individuals we can draw from is basically
already utilized.  Most of the GDB global maintainers are already local
maintainers; the number of active GDB contributors who aren't already
maintainers of some part of GDB is vanishingly small.

No matter how we place more maintenance resources, they have to come
from the existing pool.  I believe that the view of global maintenance
that we proposed at the beginning of this discussion is the most
efficient way to make use of that pool, and is the most effective way
to improve the GDB development process.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer