[PATCH V2] patches: release the GIL while applying the patch

[Augie Fackler]

On Thu, Feb 22, 2018 at 09:41:01PM +0900, Yuya Nishihara wrote:
> On Thu, 22 Feb 2018 12:10:12 +0100, Boris Feld wrote:
> > # HG changeset patch
> > # User Boris Feld <boris.feld at octobus.net>
> > # Date 1517839431 -3600
> > #      Mon Feb 05 15:03:51 2018 +0100
> > # Node ID 585005c9c4901f4f94847e8637fbc58cc5b29c56
> > # Parent  0c34cb461a1ea5d3f8e1300e0b8bc16ed8fa8802
> > # EXP-Topic parallel-patching
> > # Available At https://bitbucket.org/octobus/mercurial-devel/
> > #              hg pull https://bitbucket.org/octobus/mercurial-devel/ -r 585005c9c490
> > patches: release the GIL while applying the patch
> >
> > This will allow multiple threads to apply patches at the same time.
> >
> > diff --git a/mercurial/cext/mpatch.c b/mercurial/cext/mpatch.c
> > --- a/mercurial/cext/mpatch.c
> > +++ b/mercurial/cext/mpatch.c
> > @@ -109,7 +109,9 @@ static PyObject *patches(PyObject *self,
> >             goto cleanup;
> >     }
> >     out = PyBytes_AsString(result);
> > +	Py_BEGIN_ALLOW_THREADS
> >     r = mpatch_apply(out, in, inlen, patch);
> > +	Py_END_ALLOW_THREADS
>
> Bad macro. I've inserted one more block to make clang-format happy.

Indeed. Python has a few of these, and I periodically pester some
clang-format devs at Google to try and get some progress. That might
finally have borne fruit in the last month: it sounds like eventually
we'll be able to define a macro to clang-format as ending with one of
{,;}, so the formatter will be able to do less-dumb things.

>
>       /* clang-format off */
>       {
>               Py_BEGIN_ALLOW_THREADS
>               r = mpatch_apply(out, in, inlen, patch);
>               Py_END_ALLOW_THREADS
>       }
>       /* clang-format on */
> _______________________________________________
> Mercurial-devel mailing list
> Mercurial-devel at mercurial-scm.org
> https://www.mercurial-scm.org/mailman/listinfo/mercurial-devel