patch for issue 4423

Robert Collins robertc at
Wed Oct 29 14:21:40 CDT 2014

On 30 October 2014 05:01, Mads Kiilerich <mads at> wrote:
> On 10/28/2014 09:39 AM, Robert Collins wrote:
>> Hi - resending here as apparently they can't be discussed in the bug
>> tracker. This is a driveby - it fixed it for me, but I make no claims
>> about correctness, other use cases or testing.
> Thanks. How driveby is it?

It's pretty driveby. I had about 20K patches in cpython to sort
through, and this was essential in resolving many of them.

If its conceptually interesting the to Mercurial community, then I
certainly want to make sure you can take the code - so if there is
anything you need (e.g. a DCO) I'll be most happy to do that.

As far as polishing the code, writing tests etc, I have very little
interest in that - and a long list of much more pressing issues.

> Could you contribute an updated patch that comply with
> ? (Or actually two
> patches for what seems to be two different issues, and including test
> coverage/updates for both.)

Its one issue - the code structure is such that two changes were
needed: one to the resume function, and one to the code that calls it
to handle None as the return value for the new commit... indicating no
new commit. Alternative implementations such as an exception to signal
'ok but no new commit' are also possible, but would still need two
alterations, and committing either without the other wouldn't make

> Or is it just thrown out here take-it-or-leave-it-ish so someone else who
> cares can pick it up and take it the last mile?

This :).

Robert Collins <rbtcollins at>
Distinguished Technologist
HP Converged Cloud

More information about the Mercurial-devel mailing list