| Submitter | Pierre-Yves David |
|---|---|
| Date | Feb. 23, 2016, 1:19 p.m. |
| Message ID | <61be787ee62ea5910b0e.1456233562@marginatus.alto.octopoid.net> |
| Download | mbox | patch |
| Permalink | /patch/13311/ |
| State | Accepted |
| Commit | 48e1a641765d541a4bdc3f36b69cb3f5974040af |
| Delegated to: | Matt Mackall |
| Headers | show |
Comments
On 02/23/2016 02:19 PM, Pierre-Yves David wrote: > # HG changeset patch > # User Durham Goode <durham@fb.com> > # Date 1454629084 28800 > # Thu Feb 04 15:38:04 2016 -0800 > # Branch stable > # Node ID 61be787ee62ea5910b0e55b5b5743079fbea1457 > # Parent 1bcb4f34b9f91a2e330966182f691664fbada1bc > # Available At http://hg.netv6.net/marmoute-wip/mercurial/ > # hg pull http://hg.netv6.net/marmoute-wip/mercurial/ -r 61be787ee62e > obsolete: fix n^2 marker computation behavior This is a grafting on stable of a change from durham currently on default. This is hitting people in the wild and the change is small enough for stable in my opinion.
On Tue, Feb 23, 2016 at 02:54:52PM +0100, Pierre-Yves David wrote: > > > On 02/23/2016 02:19 PM, Pierre-Yves David wrote: > ># HG changeset patch > ># User Durham Goode <durham@fb.com> > ># Date 1454629084 28800 > ># Thu Feb 04 15:38:04 2016 -0800 > ># Branch stable > ># Node ID 61be787ee62ea5910b0e55b5b5743079fbea1457 > ># Parent 1bcb4f34b9f91a2e330966182f691664fbada1bc > ># Available At http://hg.netv6.net/marmoute-wip/mercurial/ > ># hg pull http://hg.netv6.net/marmoute-wip/mercurial/ -r 61be787ee62e > >obsolete: fix n^2 marker computation behavior > > This is a grafting on stable of a change from durham currently on default. > This is hitting people in the wild and the change is small enough for stable > in my opinion. I concur, but I'll leave that to mpm (cc'd) to decide. > > -- > Pierre-Yves David > _______________________________________________ > Mercurial-devel mailing list > Mercurial-devel@mercurial-scm.org > https://www.mercurial-scm.org/mailman/listinfo/mercurial-devel
On 02/23/2016 07:05 PM, Augie Fackler wrote: > On Tue, Feb 23, 2016 at 02:54:52PM +0100, Pierre-Yves David wrote: >> >> >> On 02/23/2016 02:19 PM, Pierre-Yves David wrote: >>> # HG changeset patch >>> # User Durham Goode <durham@fb.com> >>> # Date 1454629084 28800 >>> # Thu Feb 04 15:38:04 2016 -0800 >>> # Branch stable >>> # Node ID 61be787ee62ea5910b0e55b5b5743079fbea1457 >>> # Parent 1bcb4f34b9f91a2e330966182f691664fbada1bc >>> # Available At http://hg.netv6.net/marmoute-wip/mercurial/ >>> # hg pull http://hg.netv6.net/marmoute-wip/mercurial/ -r 61be787ee62e >>> obsolete: fix n^2 marker computation behavior >> >> This is a grafting on stable of a change from durham currently on default. >> This is hitting people in the wild and the change is small enough for stable >> in my opinion. > > I concur, but I'll leave that to mpm (cc'd) to decide. Gentle ping.
On Mon, 2016-03-07 at 16:10 +0100, Pierre-Yves David wrote: > > On 02/23/2016 07:05 PM, Augie Fackler wrote: > > On Tue, Feb 23, 2016 at 02:54:52PM +0100, Pierre-Yves David wrote: > > > > > > > > > On 02/23/2016 02:19 PM, Pierre-Yves David wrote: > > > > # HG changeset patch > > > > # User Durham Goode <durham@fb.com> > > > > # Date 1454629084 28800 > > > > # Thu Feb 04 15:38:04 2016 -0800 > > > > # Branch stable > > > > # Node ID 61be787ee62ea5910b0e55b5b5743079fbea1457 > > > > # Parent 1bcb4f34b9f91a2e330966182f691664fbada1bc > > > > # Available At http://hg.netv6.net/marmoute-wip/mercurial/ > > > > # hg pull http://hg.netv6.net/marmoute-wip/mercurial/ -r > > > > 61be787ee62e > > > > obsolete: fix n^2 marker computation behavior > > > > > > This is a grafting on stable of a change from durham currently on default. > > > This is hitting people in the wild and the change is small enough for > > > stable > > > in my opinion. > > > > I concur, but I'll leave that to mpm (cc'd) to decide. > > Gentle ping. The duplicate summary caused this to be swallowed by my "have I already dealt with this?" tools. For future reference, I think avoidable quadratic (or worse!) behavior is fair to consider an outright bug and appropriate for stable, especially if we see it manifest in 100s+ overhead. What I don't like to see on stable is performance tweaking with small constant factors. A 2x speed-up is awesome.. but not worth risking regression on stable for. I've manually grafted this here, thanks.
Patch
diff --git a/mercurial/obsolete.py b/mercurial/obsolete.py --- a/mercurial/obsolete.py +++ b/mercurial/obsolete.py @@ -1223,10 +1223,11 @@ def createmarkers(repo, relations, flag= metadata = {} if 'user' not in metadata: metadata['user'] = repo.ui.username() tr = repo.transaction('add-obsolescence-marker') try: + markerargs = [] for rel in relations: prec = rel[0] sucs = rel[1] localmetadata = metadata.copy() if 2 < len(rel): @@ -1241,10 +1242,19 @@ def createmarkers(repo, relations, flag= npare = None if not nsucs: npare = tuple(p.node() for p in prec.parents()) if nprec in nsucs: raise error.Abort("changeset %s cannot obsolete itself" % prec) + + # Creating the marker causes the hidden cache to become invalid, + # which causes recomputation when we ask for prec.parents() above. + # Resulting in n^2 behavior. So let's prepare all of the args + # first, then create the markers. + markerargs.append((nprec, nsucs, npare, localmetadata)) + + for args in markerargs: + nprec, nsucs, npare, localmetadata = args repo.obsstore.create(tr, nprec, nsucs, flag, parents=npare, date=date, metadata=localmetadata) repo.filteredrevcache.clear() tr.close() finally: