How I Patch Piloted lately and some tricks about merge proposals

[Martin Pool]

On 12 February 2010 12:47, Stephen J. Turnbull <stephen at xemacs.org> wrote:
> Martin Pool writes:
>
>  >  * patch is just so wrong that there is no real value in it
>
> There is no such thing.  Any patch is a *strong* signal of interest by
> a potential contributor (I may be projecting since I post a lot but
> haven't submitted any code yet :-).
>
> That's not really the patch pilot's problem, and the patch should be
> rejected.  But if you're serious about recruiting contributors,
> *somebody* should be following up with such folks to (a) keep them in
> the fold and (b) hopefully improve the quality of future submissions.

The point of the project is to tip the balance much more strongly
towards getting patches in, even if that means the pilot does 90+% of
the work.  However there is a limit (at least in a thought experiment)
at which the patch is so small and the bug so large and/or unimportant
that it is not economical to just finish it off.  It is of course
still worth following up with people and making it clear what they
should do next.

It's not a question of whether working on the bug is good; that's true
for all valid bugs.  It's whether it is better than other things we
could be doing.

Maybe you'd like to follow up with some of these people?

>  > One interesting case is where someone has posted a patch that
>  > reproduces the failure.  Arguably we should turn this into an xfail
>  > test.  However I think that's not worthwhile unless we're then going
>  > to actually fix the bug.

... I meant to say "fix the bug right now."  If we are never going to
fix it then yes, it should be marked as such.

> If it's a bug, and you're definitely not going to fix it, turn the
> issue into a bug and label it *wontfix*.  If you might fix it, you
> should add the test.  Doing otherwise looks like you're sweeping it
> under the rug.

After some experimentation with xfail I actually don't think having
the test in the tree is much better than just having the bug in the
database, unless you're actively working on fixing it.

If you would like to take some patches in this category from the url I
posted previously and turn them into xfail tests we will take those
patches though.

-- 
Martin <http://launchpad.net/~mbp/>