`hg status' crashes on newly created repository

[veedeehjay@…]

this happens with mercurial@3.3.3 under OSX 10.10.3. the following should reproducibly crash if the hggit' is requested in ~/.hgrc

mkdir new
cd new
hg init
hg status

with the message:

** unknown exception encountered, please report by visiting
** http://mercurial.selenic.com/wiki/BugTracker
** Python 2.7.9 (default, Dec 13 2014, 15:13:49) [GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.56)]
** Mercurial Distributed SCM (version 3.3.3)
** Extensions loaded: convert, rebase, fetch, graphlog, extdiff, hgk, color, hgsubversion, hggit, churn, pager, strip, mq, transplant
Traceback (most recent call last):
  File "/opt/local/bin/hg", line 43, in <module>
...
LOTS MORE
...
 File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/mercurial/scmutil.py", line 1105, in finddirs
    pos = path.rfind('/')
AttributeError: 'tuple' object has no attribute 'rfind'

two observations:

1. without the hggit extension the problem goes away here

2. using the mercurial-devel instead of themercurial port also makes the problem go away even in the presence of hggit. whether it's a problem of mercurial or the py-hggit port I canot say...

[mf2k (Frank Schima)]

The Priority field is for use by Macports team members only.

[veedeehjay@…]

apologies: "observation 2" in previous post is WRONG: I had the hggit extension still deactivated in ~/.hgrc when doing the test with `mercurial-devel'. so in fact both, mercurial +and+ mercurial-devel crash as desribed above if the hggit extension is requested. it might therefore be a problem of that extension, after all?

I also note, that `hg stat' works just fine with my old, existing repos. only the newly created ones are affected.

[seanfarley (Sean Farley)]

Replying to veedeehjay@…:

apologies: "observation 2" in previous post is WRONG: I had the hggit extension still deactivated in ~/.hgrc when doing the test with `mercurial-devel'. so in fact both, mercurial +and+ mercurial-devel crash as desribed above if the hggit extension is requested. it might therefore be a problem of that extension, after all?

I also note, that `hg stat' works just fine with my old, existing repos. only the newly created ones are affected.

Yes, the problem is hg-git. I use mercurial-devel with py27-hggit-devel. In fact, hg-git add quite a bit of overhead so I disable it globally and only enable it on repos that need it in $PROJECT/.hg/hgrc. Also, I'm updating all these ports today so that might help as well.

[veedeehjay@…]

Replying to sean@…:

Replying to veedeehjay@…:

apologies: "observation 2" in previous post is WRONG: I had the hggit extension still deactivated in ~/.hgrc when doing the test with `mercurial-devel'. so in fact both, mercurial +and+ mercurial-devel crash as desribed above if the hggit extension is requested. it might therefore be a problem of that extension, after all?

I also note, that `hg stat' works just fine with my old, existing repos. only the newly created ones are affected.

Yes, the problem is hg-git. I use mercurial-devel with py27-hggit-devel. In fact, hg-git add quite a bit of overhead so I disable it globally and only enable it on repos that need it in $PROJECT/.hg/hgrc. Also, I'm updating all these ports today so that might help as well.

I see. thanks for clarifying. and I will follow suit. in any case it seems a bug in hggit, then -- is it a known one or should it be reported to the mercurial guys (or whoever is behind hggit)?

[veedeehjay@…]

Replying to sean@…:

Replying to veedeehjay@…:

apologies: "observation 2" in previous post is WRONG: I had the hggit extension still deactivated in ~/.hgrc when doing the test with `mercurial-devel'. so in fact both, mercurial +and+ mercurial-devel crash as desribed above if the hggit extension is requested. it might therefore be a problem of that extension, after all?

I also note, that `hg stat' works just fine with my old, existing repos. only the newly created ones are affected.

Yes, the problem is hg-git. I use mercurial-devel with py27-hggit-devel. In fact, hg-git add quite a bit of overhead so I disable it globally and only enable it on repos that need it in $PROJECT/.hg/hgrc. Also, I'm updating all these ports today so that might help as well.

I see. thanks for clarifying. and I will follow suit. in any case it seems a bug in hggit, then -- is it a known one or should it be reported to the mercurial guys (or whoever is behind hggit)?

[seanfarley (Sean Farley)]

The bug has already been fixed ... I just need to update :-)

[seanfarley (Sean Farley)]

Should be fix now.