dirstate: add __contains__ and make __getitem__ more useful dirstate.state(f) == '?' -> f not in dirstate dirstate.state(f) -> dirstate[f]

dirstate: make wjoin function private

dirstate: break update into separate functions

dirstate: use True and false for _dirty

test-alias: Removed fallback to parentui, no longer needed since 10afa3fab6b4

merge with crew-stable

Make [defaults] in .hg/hgrc work.

fix bogus close spotted by pychecker (no close() in global scope)

Remove unused pprint import and the commented line which used it before.

Use format string for lockname again (was changed by 3e25a6eb5c9a)

Backout ad09ce1d393c and replace ''' with """ to make some highlighting happy. I have tried Debian's default emacs and the current CVS version. Default emacs doesn't have highlighting enabled (and being emacs-illiterate I don't know how to enable it) and the CVS emacs' Python highlighting has no problems with ' characters here.

removed trailing whitespace

fix unused variables reported by pychecker

churn: get current terminal width if possible

merge with crew-stable

dirstate.invalidate: avoid rebuilding _map Since hasattr will call __getattr__, the call to hasattr(self, '_dirs') will end up reparsing the dirstate file.

add dirstate._dirtypl variable Theoretically, it's possible to forget modified dirstate parents by doing: dirstate.invalidate() dirstate.setparents(p1, p2) dirstate._map The final access to _map should call _read(), which will unconditionally overwrite dirstate._pl. This doesn't actually happen right now because invalidate accidentally ends up rebuilding dirstate._map.

archive: delay extraction of file revisions This allows us to look only at the filelogs we're interested in, providing a nice speedup if we're archiving only part of a repository.

help: avoid traceback if an extension has only debug commands

convert/subversion: rehandle the no-tags case

Print meaningful error message if os.symlink fails

lock.py: cache hostname, but not pid, in case we fork

convert/subversion: work around memory leak in svn's python bindings The svn.ra.get_log wrapper attaches the hash of changed paths for every log entry to a global memory pool, so memory consumption increases rapidly, with no way to free it. Our workaround is to call this function in a child process, and feed its results back over a pipe. The memory consumption of the child still grows huge (hundreds of megabytes), but at least it goes away once the reading-the-log phase is done.

Automated merge with http://hg.intevation.org/mercurial/crew

convert/subversion: Simplify callback.

merge with crew-stable

Simplify update.

Better fix for issue 622 than we had in c4dd58af0fc8.

convert/subversion: reduce memory usage by filtering early This makes up to a 50% difference in memory consumption.

Automated merge with http://hg.intevation.org/mercurial/crew

convert/subversion: get converter working against plain HTTP.

Simplify update.

Better fix for issue 622 than we had in c4dd58af0fc8.

merge with main

doc: typo, -x bit

merge with crew-stable

fix test-tag on Solaris

tag: handle .hgtags and .hg/localtags with missing final newline (issue 601) This also fixes an asymmetry bug in which we called the pretag hook if we were going to create a local tag, but didn't call the tag hook afterwards.

issue 622: pull/unbundle -u updates to default branch if repo was empty

mq: autodetect an existing git patch during qrefresh (issue 491)

replace .hgtags instead of appending to it when doing a raw commit

handle nonexistent .hgtags in raw _tag

convert/subversion.py: str.rsplit is not available in Python 2.3

Automated merge with http://hg.intevation.org/mercurial/crew

convert/subversion.py: fix bad assumptions about SVN path naming The SVN converter assumed that the trunk and branches paths were fixed, and immediately under the base of the SVN URL. Fix the second assumption, and allow the trunk and branches paths to be reconfigured.

convert/subversion.py: escape quotes to make code readable in Emacs

patch: patches should be read and written in binary mode when possible.

patch: fix normalized paths separators.

posixfile_nt: '+' was understood as read mode instead of update.

Update test output to reflect small changes in patch chattiness.

patch.py: re-add the ability to use an external patch program This is now invoked by default only if ui.patch is set. Otherwise, we use our built-in patch. If that fails because it can't find any valid hunks, we'll fall back to trying the external patch command.

patch.py: don't mark files as changed unless they have actually been changed

patch.py: fix some incorrect uses of _() for i18n

Add Chris Mason's mpatch library. The original repo is http://oss.oracle.com/mercurial/mason/mpatch

merge with crew-stable

convert: manually set encoding to UTF-8 Right now the backends assume we're working in UTF-8 and there's no good way to ignore the encoding when adding revisions.

hg parents: don't pass an OS-specific path to repo.filectx The only reason to use the matcher code instead of calling util.canonpath directly is to be able to use "path:canonical/path/to/file" patterns.

merge with main

contexts: improve filectx eq test

Merge with -stable

merge: fix unnecessary rename merges on linear update (issue631) If one side's revision is identical to the ancestor, we skip the rest of the copy detection logic.

tag: handle .hgtags and .hg/localtags with missing final newline (issue 601) This also fixes an asymmetry bug in which we called the pretag hook if we were going to create a local tag, but didn't call the tag hook afterwards.

issue 622: pull/unbundle -u updates to default branch if repo was empty

mq: autodetect an existing git patch during qrefresh (issue 491)

patchbomb: add --desc, to specify a file containing a series description

merge with crew-stable

convert: Use clone's behaviour for the default destionation name. Adjusted the documentation for this and for the fact that the destionation is created even if a name is specified.

Merge with crew-stable.

test-symlink-basic: require symlinks availability

Add hghave utility and run-tests.py support. hghave detects system features like symlinks availability at run-time. Tests can be skipped by starting them with: "$TESTDIR/hghave" symlink || exit 80 The 80 exit status triggers hghave output handling by run-tests.py. Also, tests output can be locally patched on the fly to match reference output.

run-tests: extract correct status when script terminates with exit

Merge with -stable

diff: correctly handle combinations of whitespace options

hg grep: handle re.compile errors & update tests/test-grep

typo fix

Re: mercurial.el and hg-version-other-window hg-version-other window is unimplemented in mercurial.el. The version below works for me, YMMV.

Use isinstance instead of type == type

convert: record the source revision in the changelog

Use wsgi.url_scheme instead of ad-hoc CGI checks. Instead of each place in hgweb_mod that needs to check for SSL or get the protocol scheme (http vs. https) doing it ad-hoc, make them just look at the wsgi.url_scheme which because of previous patches is now always set correctly.

cosmetics

Make hg serve set the wsgi.url_scheme property correctly. This fixes the bug where hg serve with SSL will always detect the wrong urlbase for templates. This is also part of unifying the CGI interface and hg serve interface to both use standard wsgi.

Handle CGI SSL detection via HTTPS environment better. Some servers send on/off, 0/1, yes/no, and may be upper or lower case. This fix will handle all of those permutations. It was inspired by the detection done in in some other wsgi python web applications I looked at.

merge with crew-stable

archive: make the %r escape work.

serve: respect settings from .hg/hgrc create_server was looking only at the root ui object, ignoring any settings from .hg/hgrc. To keep respecting command-line arguments, commands.serve must also call repo.ui.setconfig.

Fix inconsistency for the stream_out capability in hgweb During some experiments of mine, the uncompressed cloning could not be enabled for hgweb.cgi nor hgwebdir.cgi though the server claimed to be stream_out capable. The only solution was to enable it using the user's .hgrc file. This solution is not acceptable when publishing the repos through an HTTP server because the CGI runs as a www dedicated user whose's home hgrc file may not be accessible to users publishing their repos through their userdir. For such cases we could end up with this typical debug output: hg --debug clone --uncompressed http://server/hg/project destination directory: project sending capabilities command capabilities: lookup changegroupsubset stream=1 unbundle=HG10GZ,HG10BZ,HG10UN sending stream_out command abort: operation forbidden by server The error lies in the fact the hgweb object defines new accessors to the repo configuration that trust things by default (untrusted=True) but the streamclone:stream_out function uses the usual accessors to the repo.ui object, which do not trust by default (untrusted=False) Fix this inconsistency, adding a new parameter to the stream_out function. hgweb then forces a "trust by default" behavior.

improve warning for hg add foo; hg mv foo bar

util._hgexecutable: default to 'hg' This should make hooks work even if util.set_hgexecutable is not called (e.g. hgweb.cgi). Fixes issue625.

archive: add symlink support

archive: use util.opener when archiving files. This simplifies the code a bit and makes it easier to support symlinks.

localrepo.wwrite: use opener.symlink This makes us fall back to regular files if symlinks are not supported.

Add symlink method to util.opener. Differences from os.symlink: - the symlink name is relative to the opener base directory - if a file with that name already exists, it's removed - if necessary, parent directories are created - if the system (OS or filesystem) doesn't support symlinks, a regular file is created. Its contents are the symlink target.

turn util.opener into a class

Make {urlbase} work in templates when https is used.

replace .hgtags instead of appending to it when doing a raw commit

handle nonexistent .hgtags in raw _tag

Export extra in _tag so convert can set the branch of a tag

archive: abort on empty repository. Fixes #624.

mq: improve qclone error handling when patch directory is not a repository.

Fix test-debugcomplete for new --certificate option

Add SSL support to hg serve, activated via --certificate option

Correct inadvertent line ending change.

Fix for win32text corrupting files that have CRLF line endings. (issue302) Also generate a warning when encountering this condition. From Shun-ichi Goto in attachment win32text-patch3.txt for issue 302.

Add --line-number option to hg annotate (issue506) Line numbers are separated from the last annotation element with a colon, like with many other commands, e.g. (hg) grep or compiler errors. Idea and tests by FUJIWARA Katsunori.

Allow filectx.annotate to return the line number of first appearance.

Check for conflict markers in hgmerge (issue471)

merge with crew-stable

Fixed double slash for doc install path introduced by using $DESTDIR (issue620) $MANDIR already is an absolute path so the extra / isn't needed.

hg log: Move filtering implicit parents to own method and use it in templater. Extended test repo in test-command-template to contain changeset to test this.

Strip whitespace from changeset description in changeset_templater. changeset_printer already does this, too.

Removed deprecated hg:// and old-http:// protocols (issue406)

hgwebdir: check for repo foo before browsing subdirectories of foo/

Merge with crew

hgweb: browse subdirectories before checking whether parent directory is also a repository

Use a prefix for debuginstall tempfiles. This way you know where the files came from if they don't get removed.

python 2.3 does not have sorted

Sort changelog extra dict to avoid possible nondeterminism

Last change broke new-style URLs

hgwebdir: do not walk up the given path looking for a repository. It is there or it isn't.

merge with crew-stable