Probability of acceptance of pull requests

The title is a much more cynical take on the data than I have looking at it. I have some familiarity and experience with Django, one of the projects that has a pretty low pull request acceptance rate from his sample. At 29% I was impressed that many pull requests were accepted, frankly! I think the signal to noise is pretty high on pulls to high profile projects.

It's easy, and common I think, to throw a small change into a pull request and fire it off without confirming it's desired or within the objectives of the maintainers. I've done this at least once w/ Django personally. My thought process was "This may be undesired, but if it is desirable the chances of it making it in without bikeshedding in a mailing list is way higher if I just lead with a pull request."

In my case it wasn't desired (I think it was a settings.py var that allowed users to configure auto-import of models in the shell, dropped cause it's implicit over explicit and adds to settings.py).

Add to that all of the pull requests that are effectively proof of concepts because the implementation fails to take some major stuff into account, and an acceptance rate of almost a third starts to sounds pretty good to me.

Re: Probability of acceptance of pull requests

#6

Post by neil_s » Wed, May 29, 2013, 12:40 AM UTC

The vertical labels on the charts reduce readability greatly! Perhaps angle them a bit and sort from highest to lowest so you can just scan down the labels and know where they stand without having to look back up at the bar height. Or simply flip the axis.

Re: Probability of acceptance of pull requests

#7

Post by zalew » Wed, May 29, 2013, 12:45 AM UTC

great analysis. I thought big projects have a lower acceptance rate than smaller ones in general, but when you take a look at django vs flask to rails vs sinatra, it doesn't seem to be the case.

one nitpick: next time rotate the diagrams +90' please :)

Re: Probability of acceptance of pull requests

#8

Post by zalew » Wed, May 29, 2013, 12:51 AM UTC

famousactress wrote: post #5

The title is a much more cynical take on the data than I have looking at it. I have some familiarity and experience with Django, one of the projects that has a pretty low pull request acceptance rate from his sample. At 29% I was impressed that many pull requests were accepted, frankly! I think the signal to noise is pretty high on pulls to high profile projects. It's easy, and common I think, to throw a small change…

well, I guess adding anything to settings.py will get red flags instantly because everyone is sick of settings, that is common knowledge when you follow django-dev and con vids.

my almost first contribution to django was recently hijacked by a better one, bummer :)

Re: Probability of acceptance of pull requests

#9

Post by jtchang » Wed, May 29, 2013, 12:52 AM UTC

What is a CLA and why would a project want one / not want one? They seem like another hurdle for contributors?

Re: Probability of acceptance of pull requests

#10

Post by gresrun » Wed, May 29, 2013, 12:53 AM UTC

The curious question is why certain projects have significantly lower pull-rates; I doubt the quality of the pull requests vary that widely across projects.

Is it that project leadership has a clear vision for a product and feel pull requests are a distraction? If so, share the vision and enlist willing developers to help achieve that vision.

Is it that project leadership has a high-standard for the codebase? If so, explicitly state the standard and provide constructive feedback to pulls that don't quite measure up.

I guess my point is: project leadership has the ability to leverage a tremendous pool of talent that is interested in their project using the pull request mechanism. It does take effort and planning but the net result is so worth it to all parties involved.