Rebase now doesn't make assumptions about which format has been
used for a mq patch (git or normal).
Before finalizing a patch it keeps track of the original format,
by reading its header, and then restores the format when reimporting
it.
This way it also allows for having mixed styles.
Note: this version corrects a failure in the test
If the user created the clone target directory before running
the clone command, only cleanup the .hg/ repository when errors
occur. Leave the empty target directory in place.
inotify is smart enough to notify you about any changes in a
directory, even if you only watch the directory, and none if its
contents: the recursive add_watch I added was unnecessary.
The only thing that matters here is the recursive status update on
directory deletion.
And scan() has to be called _before_ the deferred call is registered.
(race condition: depending on the times, the previous patch could
apparently fail on the provided test. It's not the case anymore.)
This fixes issue 1569. hg churn <path> now returns only the number
of lines changed in the path, if the path is specified by filtering
files through a match filter at the changeset level. test-churn
has been updated to take care of this issue.
Use subprocess to emulate Popen4 if available - similar to how it is done in
util.py.
Using popen2 under python 2.6 gives
DeprecationWarning: The popen2 module is deprecated. Use the subprocess module.
- add writepending to flush delayed writes to separate file
- add support in hooks for lazy evaluation of callable parameters
- add HG_PENDING to pretxn hooks
- call writepending if hook is used
- pass repo root to hook environment
- if HG_PENDING = repo root, we're in pretxn hook
- read pending data to make pending changesets visible
- filter HG_PENDING in tests/printenv.py
The merge tool selecting algorithm is hardcoded to look for hgmerge in $PATH
before it falls back to use internal merge. This fixes the test for this
fallback to be tolerant to existing hgmerges.