Submitter | Mads Kiilerich |
---|---|
Date | Dec. 16, 2012, 10:34 p.m. |
Message ID | <d8ae1e3965a185b2dda6.1355697255@localhost6.localdomain6> |
Download | mbox | patch |
Permalink | /patch/144/ |
State | Accepted |
Commit | 5965997b70235e768c4244454afae5ca86b3caaa |
Headers | show |
Comments
On Sun, Dec 16, 2012 at 2:34 PM, Mads Kiilerich <mads at kiilerich.com> wrote: > convert: process splicemap in sorted order > This doesn't have any visible consequences? -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://selenic.com/pipermail/mercurial-devel/attachments/20121217/b9848f34/attachment.html>
Bryan O'Sullivan wrote, On 12/17/2012 10:09 PM: > On Sun, Dec 16, 2012 at 2:34 PM, Mads Kiilerich <mads at kiilerich.com > <mailto:mads at kiilerich.com>> wrote: > > convert: process splicemap in sorted order > > > This doesn't have any visible consequences? Correct, not with the default hash seed. Some tests (test-splicemap.t IIRC) would sometimes fail when running with random seed - either because of different output ordering or because of assumptions in the test code. That is actually the case for most of the patches in this series. Enumerating the failures I have seen and how they were fixed is not feasible. /Mads
Patch
diff --git a/hgext/convert/convcmd.py b/hgext/convert/convcmd.py --- a/hgext/convert/convcmd.py +++ b/hgext/convert/convcmd.py @@ -147,7 +147,7 @@ map contains valid revision identifiers and merge the new links in the source graph. """ - for c in splicemap: + for c in sorted(splicemap): if c not in parents: if not self.dest.hascommit(self.map.get(c, c)): # Could be in source but not converted during this run