Re: Maintainers and contributors

[John Wiegley]

>>>>> Eli Zaretskii <address@hidden> writes:

> I would consider it unfair to hand off such a "discussion" -- unfair to the
> person to whom I'm handing it off. If someone wants that crown of thorns,
> they should volunteer.

I'm happy to have that crown of thorns, as you call it. It would make you more
productive, Eli, by not draining away your energy in these sorts of disputes.
On the other hand, I have a personal need to improve my ability to negotiate
these types of scenarios (outside of Emacs), so it gives me a chance to hone
my fledgling skills.

I think Artur's suggestion of shifting the discussion after it bounces back
twice on the same point has merit. Not in the shadowy-cabal-backroom sort of
way that David is opposed to, but in the "We hear your concern and are going
to escalate its importance" sense. If I were arguing a point with a single
maintainer, and that maintainer said, "Clearly this issue is of importance to
you, I'm going to bring another maintainer into this discussion", I would feel
very listened to.

As it stands, the falling out with Taylan was not entirely technical. I've
spoken to Taylan on IRC, and he is actually a very reasonable fellow. Mainly,
there was a difference between his desire, and his position, that we missed:

  Desire: Avoid security vulnerabilities in his code.

  Position: `shell-quote-argument' violates this desire, and should not be
          used. Since emacs-devel probably can't fix `shell-quote-argument'
          today, rewrite it until it is fixed.

Had the discussion been about this desire, we could have talked about whether
he should bother worrying about security in the context of Emacs, since we
generally don't put much focus there. Eli did start to mention this, but I
think it was lost in the storm, or seen as a dodge.

Because the `shell-quote-argument' position was stated early in the bug
thread, the discussion devolved into a "hard bargaining" scenario, where
Taylan could not accept using `shell-quote-argument' as it stood, and we could
not accept his re-implementing it. From that moment on there was really no
agreement possible, not without sacrifice. This is when things started to get
nasty, because the submitter thought we were completely ignoring his primary
issue.

Here's what our side looked like:

    Desire: Make Emacs as easy to maintain as possible.

    Position: Re-implementing `shell-quote-argument' is unnecessary; if it
            has problems, we should fix it, rather than increasing our code
            surface.

Was there a solution to resolve both of these desires? I bet you there was.
Was there one to unite the two positions? Doubtful; at least, not in a manner
satisfying to both parties.

I'm not asking the maintainers on emacs-devel to become negotiators who must
worry about layers of meaning: only that Artur's suggestion of escalating
disputatious issues -- so we can step back together and reassess the needs of
the submitter -- could have made it possible to avoid all that was lost in the
past week.

John