On Tue, Mar 26, 2013 at 5:42 AM, Aaron Meurer <[email protected]> wrote:
> We currently have 92 open pull requests, and that number is only going
> to get much worse in April if we are accepted for GSoC.  Working on
> the GSoC application is high priority now, but we should also focus on
> clearing the pull requests a bit.
>
> For those open pull requests that are marked as WIP, we need to
> determine exactly what work still needs to be done, and more
> importantly, if it is essential or not that the pull request not be
> merged until that work is completed. It's tempting to mark a pull
> request as WIP until you have done all the work you want to do, but if
> the work that is done so far is OK, then it should be merged, and new
> work put in a new pull request.
>
> I think there are several pull requests that are fine to go. They just
> need a final review (or a review at all). My two most recent pull
> requests (1916 and 1876) fall into this category.
>
> For those that really do have work to do, add a TODO to the pull
> request description (or a comment if you don't have push access). It's
> nice to use the GitHub Markdown checkbox lists, which work like
>
> - [ ] item1
> - [ ] item2
>
> Then the items can be checked off as they are completed, and once
> everything is checked, you will know that it is done.
>
> Better yet, send a pull request to the person who made the pull
> request fixing the issues. Just fix the issues in your branch, and go
> like you are making a pull request to SymPy, but then change it to go
> against the person's branch.  If GitHub doesn't show the person's name
> in the list, go like you are sending a pull request to someone else
> and then change the name in the url to the one you want.
>
> And even though we do want to clear the pull request queue, please
> don't merge any pull request unless it has a passing Travis review, or
> if Travis is acting up, a passing SymPy Bot review. Clearing the pull
> request queue is important, but keeping tests passing in master is
> more important.

I would suggest to adopt the ipython approach:

* create a policy like:
https://github.com/ipython/ipython/wiki/Dev:-Closing-pull-requests

* find pull requests that are waiting for a response of the original
author for more than a month (or require large redesign), close them
and post a link to our policy

* create a new issue with a link to the PR

Here is an example of my own PR, that I simply don't have time to finish up:

https://github.com/ipython/ipython/pull/2504

That is the only reasonable way in my opinion. The PR queue should be
for pull requests that are being worked on.
There is absolutely nothing wrong with closing a pull request that
nobody has time to finish.

I went ahead and created our own policy page here:

https://github.com/sympy/sympy/wiki/Dev:-Closing-pull-requests

Feel free to modify it/update it. Once we all agree with the text,
I'll go over old PRs and close them according to the policy (plus
create an issue with it).

Ondrej

-- 
You received this message because you are subscribed to the Google Groups 
"sympy" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sympy?hl=en-US.
For more options, visit https://groups.google.com/groups/opt_out.


Reply via email to