a simple request: apply all outstanding patches.

[Rob Landley]

On Wednesday 17 May 2006 9:15 am, Natanael Copa wrote:
> On Wed, 2006-05-17 at 08:39 -0400, Robert P. J. Day wrote:
> > i suspect patch submission works much the same way.  as long as there
> > are few submissions, no problem.  but if you start to overwhelm the
> > maintainer and patches start taking longer to be processed, submitters
> > assume their patch got lost so they re-submit, which puts more work on
> > the maintainer, which causes him to fall further behind, which
> > inspires submitters to re-submit again ... you get the idea.
>
> Yes. Maybe we should stop submitting patches, except those who actually
> help rob forward, until he get on top of the situation again. Then we
> can discuss how we can help him _how_ to get the patches commited in a
> timely manner.

Since the start of this month (starting at svn 14976) I've committed 66 
patches.  In the entire previous month (svn 14722) I committed 55 patches.  I 
don't think the rate at which I commit patches is the fundamental problem. :)

The problem is the rate of busybox development is increasing.  For reference, 
Erik called for help on April 14, 2005: 
http://busybox.net/lists/busybox/2005-April/014158.html

Between January 1, 2005 (svn 9633) and April 14, 2005 (svn 10102) a grand 
total of 36 patches were checked in.  By everybody, combined.  Now admittedly 
part of that involved a cvs to svn migration, but that's still a three and a 
half month period with about half as much development going into the tree 
than I've personally committed in the last 17 days.  (And I'm not the only 
committer.)

I started committing other people's patches with svn 10201, and had to revert 
some of them almost immediately, but I kept at it because nobody else stepped 
up to do it.  (Vapier checked in a lot of stuff about this time, but you'll 
notice it was almost entirely his own development work.  The e2fs port, 
adding ed to the patches directory, adding stat, whitespace fixes, 
modifications to existing applets that weren't attributed to someone else in 
the checkin comment, etc.  He wasn't reviewing and checking in other people's 
patches, which isn't fun, isn't easy, and isn't quick.)

I'm not an ideal maintainer.  Not even close.  I'm not good at the social 
aspects of it (I started out by having a personality conflict with Vladimir, 
and I suspect I've managed to annoy everyone on this list at least once by 
now), I drop patches on the floor, I post too much and outright _flood_ the 
darn mailing list (hey, I type fast), I haven't got enough time to spend on 
patch review (more energy than time, it's _exhausting_ reading this stuff 
with a fine tooth comb), I'm not patient enough, there are areas (like 
makefiles) where I really don't know what I'm doing...

Unfortunately, the reason I'm doing it is there isn't anybody else.  I put a 
lot of time in this, I care deeply about what's best for the project, and I'm 
doing the best job I can.

If Erik wanted the job back I'd dump it in his lap in a heartbeat, but I'm 
pretty sure the reason he stopped having time for it is he only has time for 
two projects.  When buildroot forked off of uClibc, something had to give, 
and it was BusyBox.  And that means your stuck with me...

Rob
-- 
Never bet against the cheap plastic solution.