[hibernate-dev] Retrospective on "Pull Requests": a waste of time?

[Hardy Ferentschik]

On 10 Jan 2013, at 2:04 PM, Sanne Grinovero <sanne at hibernate.org> wrote:

> - your metadata API is not complete, yet I merged two chunks of
> progress, as I considered them solidly good stuff. You don't disagree
> do you?

No, of course not, but how does this example fit into this? If you had told me
that you don't want to merge until everything is complete I would have closed the
pull request and continued working. However, in this case we agreed to merge,
because

a.) it was as you said already quite solid
b.) the whole issue is not completed yet, so I can integrate your suggestions in a follow up
     pull request (mind you same issue!)
c.) there were this check style changes coming up and I rather had Davide changing my
     quite extensive change set in one go, instead of me going through some painful 
     rebase.

I find c.) in this case a particulate string point. It avoided a lot of unnecessary work

> - I'm upgrading Infinispan and get side-tracked by a test which is
> not cleaning up properly, so I fix the cleanup as I need it, but
> technically this already is a costly context switch for me. You then
> identify more things that could be improved on that same test - and
> rightfully

So we agreed that some improvement could be made. The change was a no brainer.
Even you have to admit this now. It literally changed two lines. 

The problem here was not the fact that we are to nitty gritty with reviews or don't
process pull request fast enough. It was a plain miscommunication issue. 

You did not understand what I wanted or how it could be achieved in a simple way
and I did not understand why you could not just make this minor adjustment and be
done with it.  Nothing wrong with the process.

> I guess you were not aware
> that I wasn't working on that test, and hope you see the point now?

No, not at all. Obviously you worked on the test, since you made a change and included it
into the pull request. Are you saying that this change ended up accidentally in the 
pull request?

--Hardy