Re: improving our contributing tools and workflow

[Joseph Rushton Wakeling]

On 26/09/13 14:52, Phil Holmes wrote:
> I thought I made this clear - I was repeating something Graham said to me on a
> number of occasions.  He would argue it was realistic, not pessimistic. You have
> to be aware of the fact that, simply by working hard on a problem does not
> guarantee that the effort expended will be rewarded.
>
> Here's a direct quote from him - clearly you don't fall into the category of new
> contributor, but the warning still applies:
>
> "We've had bad experiences where a helpful and enthusiastic new
> contributor misunderstood the instructions, ran off and did 5 hours of
> work instead of 10 minutes, and none of the main developers wanted to
> take the time to deal with the results of the 5-hour work, so the
> whole thing was wasted. (literally wasted, as in "the project would
> have received more benefit from the 10-minute job instead of the
> 5-hour work")"
>
> Check the results of the grand regression test review.

This risks becoming another corrosive discussion, so please understand that what 
I say next is not intended as an attack on anyone here and is meant in a spirit 
of hope for Lilypond's prosperous future.

There is another possible response to such a situation, and it's: "Oh wow, this 
person put a load of work in, they're obviously really committed and 
enthusiastic.  OK, let's use these problems with what they've done as an 
opportunity to educate them better about how Lilypond works and how to avoid 
these kinds of problem in the future, and make them feel that we really value 
the time they've put in and want to repay them in kind."

Now, I'm not assuming that no one has ever done this.  I rather imagine it's 
been tried and that the resulting workload (probably mostly Graham's) has been 
overwhelming and that in fact there is no guarantee that it pays off in terms of 
another long-term contributor -- so people have been discouraged from this 
approach by hard experience.  But I still think that it's possible to approach 
contributors with enthusiastic caution rather than lowered expectations, which 
are demoralizing for everyone.

FWIW I think automated testing of pull requests is helpful here because test 
failures are impersonal and encourage the contributor to pro-actively sort out 
the problems in their code without having to be told -- there's not the same 
sense of personal rejection.