Re: [Gdbheads] Re: Feb's patch resolution rate

[Jim Blandy]

Andrew Cagney <address@hidden> writes:
> > On Tue, 23 Mar 2004 16:31:52 -0500, Andrew Cagney <address@hidden> said:
> >
> >>> BTW, did you ever get round to doing an analysis on who was reviewing
> >>> the symbol table patches.
> > I haven't done a formal analysis, no
> 
> Here, for instance, is a graph illustrating how many patches over and
> above Jim, elena reviewed '03:
> 
> 01: ------------------
> 02: ++++------------------------------
> 03: +----
> 04: +----------
> 05: +-------------
> 06: -----------
> 07: ++-------------
> 08: --+
> 09: ++++----------
> 10: +++-------------
> 11: +++-----
> 12: +++++----
> 
> +: Jim; -: Elena (Hmm, what happened in August?), and here's who
> reviewed your patches:
> 
> 01: ----
> 02: ----------
> 03: --
> 04: --
> 05: -----
> 06: ---
> 07:
> 08:
> 09: +
> 10:
> 11:
> 12:
> 
> Hmm, I should have also counted who reviewed Daniels patches as I now
> suspect the results would be similar.
> 
> If anything it should be Elena and you maintaining the symbol table.


About a year ago, you called me on the phone and asked me to remove
myself as a symtab maintainer.  I declined to: I have made substantial
contributions to the code from time to time; I understand the code
well; and from time to time I am able to make review a higher priority
and contribute more than usual.  So I didn't see how it was in GDB's
interests to remove myself from the list of people able to approve
symtab patches.

I can think of many reasons to ask a maintainer to step down.  Perhaps
he objects to the addition of other maintainers; or he doesn't
understand the code; or he prevents other maintainers from
contributing; or there is a fixed limit on the number of maintainers.
But none of those hold in my case.

I didn't understand why your request made things better for GDB, so I
declined it.  I asked you several times, from several different
angles, why it was better for GDB to have fewer people able to review
patches.  I never got an answer I understood, and was left wondering
if the reasons for your request were as you had presented them.

(If someone can persuade me that the rate of contributions to GDB will
be improved by my stepping down, I'll gladly do so.  But mostly,
people have been mystified by Andrew's reasoning on this point.)



I think the big picture here is that, although different volunteers
contribute unequally --- and those who contribute more deserve the
appreciation of the community, which I think Elena has --- we don't
have enough total volunteer hours to keep our patch review up to the
standards people have come to expect from other projects.

Why don't we have more volunteer hours?  Is there anything about our
process that discourages new volunteers?  Is there anything we can do
to get more hours from the volunteers we have?

The proposal our group put together has two parts, which tackles the
problem from two different angles:

- First, allow global maintainers to approve patches anywhere.  This
  means that there are more people who can legitimately review a
  patch.  (And it's more in line with the process used by other GNU
  projects.)

- Second, provide a way to resolve disputes that reflects the general
  opinion of the group, not merely Andrew's opinion.  We've already
  posted several links to threads showing a situation which comes up
  fairly often: Andrew digs in his heels on some issue, and although
  the general consensus is against him, he wins, because he's the head
  maintainer.  This is the way our procedure works at the moment.

  One effect of this procedure over time is to make heel-digging a
  successful strategy for Andrew: if he is sufficiently stubborn, he
  always gets his way.  The other effect is to really annoy the people
  who are trying to contribute something, and make them much less
  interested in continuing to do so.

  If issues can be resolved with a vote, then Andrew is forced to
  actually persuade the other participants in the discussion, not
  merely out-stubborn them.  This will encourage Andrew to explain
  himself more, and allow people who have been able to gather the
  general opinion to their favor to actually get their work checked
  in.

  In the end, this will make GDB more friendly to volunteers, and
  attract more of their hours.