| Submitter | Matt Harbison |
|---|---|
| Date | June 30, 2015, 2:26 p.m. |
| Message ID | <b84b980cf15bdda20c47.1435674395@MATT7H-PC.attotech.com> |
| Download | mbox | patch |
| Permalink | /patch/9830/ |
| State | Accepted |
| Headers | show |
Comments
On Tue, 2015-06-30 at 10:26 -0400, Matt Harbison wrote: > # HG changeset patch > # User Matt Harbison <matt_harbison@yahoo.com> > # Date 1435588496 14400 > # Mon Jun 29 10:34:56 2015 -0400 > # Node ID b84b980cf15bdda20c474f8f5ff6b3e367164f7e > # Parent ff5172c830022b64cc5bd1bae36b2276e9dc6e5d > revset: fix a crash in parents() when 'wdir()' is in the set This one's queued for default, thanks. FYI, we've now got breakage in test-subrepo-deep-nested-change.t in this area. The calculation of changessince in archive explodes due to trying to use repo.revs("%d") on a workingctx. I've got a patch to add a %c revset.formatspec that knows how to take a context object and turn it into a rev or 'wdir()', but this isn't quite sufficient because ::wdir() doesn't work and wdir()^ has the same bug as parents().
On Tue, 30 Jun 2015 15:15:35 -0400, Matt Mackall <mpm@selenic.com> wrote: > On Tue, 2015-06-30 at 10:26 -0400, Matt Harbison wrote: >> # HG changeset patch >> # User Matt Harbison <matt_harbison@yahoo.com> >> # Date 1435588496 14400 >> # Mon Jun 29 10:34:56 2015 -0400 >> # Node ID b84b980cf15bdda20c474f8f5ff6b3e367164f7e >> # Parent ff5172c830022b64cc5bd1bae36b2276e9dc6e5d >> revset: fix a crash in parents() when 'wdir()' is in the set > > This one's queued for default, thanks. > > FYI, we've now got breakage in test-subrepo-deep-nested-change.t in this > area. The calculation of changessince in archive explodes due to trying > to use repo.revs("%d") on a workingctx. I've got a patch to add a %c > revset.formatspec that knows how to take a context object and turn it > into a rev or 'wdir()', but this isn't quite sufficient because ::wdir() > doesn't work and wdir()^ has the same bug as parents(). My plan is to finish the keyword Yuya and I have been debating, and then switch the archive code to that is possible to avoid a copy/paste. I've also got a series that fixes wdir()^, among other things. I fixed a crash in ::wdir(), but need more investigation because wdir() doesn't show in the series.
Patch
diff --git a/mercurial/revset.py b/mercurial/revset.py --- a/mercurial/revset.py +++ b/mercurial/revset.py @@ -1468,7 +1468,10 @@ def parents(repo, subset, x): ps = set() cl = repo.changelog for r in getset(repo, fullreposet(repo), x): - ps.update(cl.parentrevs(r)) + if r is None: + ps.update(p.rev() for p in repo[r].parents()) + else: + ps.update(cl.parentrevs(r)) ps -= set([node.nullrev]) return subset & ps diff --git a/tests/test-add.t b/tests/test-add.t --- a/tests/test-add.t +++ b/tests/test-add.t @@ -75,6 +75,13 @@ should fail $ hg ci -m 0 --traceback + $ hg log -r "heads(. or wdir() & file('**'))" + changeset: 0:* (glob) + tag: tip + user: test + date: Thu Jan 01 00:00:00 1970 +0000 + summary: 0 + should fail $ hg add a @@ -99,6 +106,15 @@ should fail M a ? a.orig +wdir doesn't cause a crash, and can be dynamically selected if dirty + + $ hg log -r "heads(. or wdir() & file('**'))" + changeset: 2:*+ (glob) + parent: 2:* (glob) + parent: 1:* (glob) + user: test + date: * (glob) + should fail $ hg add a