Merge with crew-stable

patch: better handling of sequence of offset patch hunks (issue1941) The built-in patch implementation applied the hunks to the wrong lines of the file if the file in the repo has been modified to skew the patch line numbers and the file contains repetitive sequences of lines.

Makefile: added instructions on how to override PREFIX

Merge with stable

win32text: do not mention hgext in warning

patch: drop unused hunk.newctrl()

test-import-eol: improve binary hunks test

diffhelpers: handle CRLF in hunk last line without EOL Note that diff generation code always ends last lines without newline with a LF, so this is a tolerance.

patch: drop eol normalization fast-path for 'lf' and 'crlf' With eolmode set to 'lf' or 'crlf' we avoided the hunk duplication and normalization by reading the input patch in text mode. Dropping this optimization simplifies code expectations for a small overhead. The change in test-mq-eol comes from a tolerance to CRLF instead of LF for last lines without newlines being broken by this revision. This tolerance was only partially supported and will be added again in a better way.

patch: fix eolmode=auto with new files If target file does not exist or has no eol, current code was normalizing eols to LF. Preserve patch file eols instead.

Merge with stable

tests highlight, keyword: load extensions with "ext ="

setup.py: keep Python 2.3 compatibility Though we only support Python 2.4 or greater, we should keep setup.py compatible with earlier versions on the syntactic level. Otherwise people will simply get a SyntaxError when trying to install Mercurial with an old version of Python. With this change, the setup.py file can be imported with Python 2.3 and we then issue a friendly error message when we detect that Python is too old.

Merge with stable

extensions help: make disabling example analogous to enabling example The example for enabling an extension in the module search path is: foo = foo can only be disabled by: foo = ! and not by: hgext.foo = ! As the examples now omit "hgext." the example hgext.bar = ! is misleading.

test-help: improve test coverage Adds a test for using the help option with the version option. Extends test coverage of commands.help_ by another 2 lines.

setup: ignore 'not importing' warnings during version detection Python will issue an ImportWarning when seeing 'import locale' if there is a locale/ directory present without a __init__.py file. The warning is silent by default, but it somehow shows up anyway on Windows when setup.py executed hg. The warning causes runcmd to panic since it sees output on stderr. This patch ignores warnings on stderr about not importing a package.

tests: load with "ext =" instead of "hgext.ext ="

Merge with stable

keyword: go into restricted mode when transplanting Expanded keyword is read into diff with transplant -b otherwise. Test this scenario. Thanks to Adam Berkes for providing an example case.

keyword: do not monkeypatch diff when in restricted mode kw_diff actually disabled restricted mode when 2 revisions were given, because it effectively disables the extension in this case. But the commands working with diff and patch need restricted mode always enabled, i.e. expansion enabled when writing to the working directory and - crucial for these commands - no expansion when reading the filelog.

merge with default

mq: qnew -f should reject merge working directories

Merge with stable

hgext: enable extensions without "hgext." prefix in help texts

cmdutil: replace sys.maxint with None as default value in loglimit Semantically, it is better to use None over any other value when there is "no value". Using maxint in this context is quite hackish, and is not forward compatible.

test-help: improve test coverage Adds a test for displaying the short list of commands with the verbose option i.e. 'hg -v help shortlist'. Extends test coverage of commands.help_ by 2 lines.

bookmarks: move parse() and current() into property definitions

bookmarks: refactor code responsible for updates of bookmarks introduce _bookmarksupdate for common parts from commitctx and addchangegroup

bookmarks: repo._bookmarkcurrent should be a propertycache

bookmarks: write() can simply access repo._bookmarks

bookmarks: turn repo._bookmarks into a propertycache

Merge with stable

hook: fix bug (reuse of variable) introduced in 872d49dd577a For binary installs, the 'name' argument would be reused as a local variable, destroying its original value. The patch fixes that, and also avoids copying sys.path when it's not necessary.

patch: implement patch.eol=auto mode EOLs in patched files are restored to their original value after patching. We use the first EOL found in the file, files with inconsistent EOLs will thus be normalized during this process.

patch: propagate eolmode down to patchfile The old code mapped the value of eolmode ('strict', 'crlf' or 'lf') to eol (None, '\r\n' or '\n') at the entry point in internalpatch. The value of eol was then used directly as the desired EOL in patchfile. We now delay the mapping and let patchfile do it instead. This allows for more complicated behavior where it does not make sense to map eolmode directly to the target EOLs.

Merge with stable

share: use defaultdest to compute unspecified destination This harmonizes with clone, and avoids creating a .hg in the current directory if the source ends with a /

Merge with stable

glog: fix "incompatible option" error message. Options like 'only_branch' should be presented to the user as '--only-branch'.

Merge with crew

convert: implement two hooks in builtin cvsps

Merge with crew

Merge with stable

doc/Makefile: Fix rst2html detection Displaying the output from the failing call to "which" didn't prevent make from doing stupid things later. We now only search for "rst2html" and fallback to "rst2html.py". If neither name is found, make will eventually abort when we try to use $(RST2HTML).

inotify: completely ignore events on the repository root They can only be events on the top directory object. Those events are meaningless for us: permission changes, repository-wide moves... And they break the assertion that all events happen in repo/*. Just ignore them. Fixes one of the failures of run-test.py --inotify test-permissions

inotify: do not rely on stat(.hg/dirstate) to invalidate our dirstate stat() is not reliable when several events happen quickly. Which means that if two hg actions occur in the same second, stat() result will not reflect the second change. And only _one_ invalidate() call was done. Also ignore the events that occur when wlock is held, since wlock release will trigger a full rescan anyway. Fixes 17 run-tests.py --inotify tests.

inotify: follow new files if they are regular or a symlink. Fixes run-tests.py --inotify symlink-basic

run-tests.py: Show paths to failing tests, .err and .out I think this is very convenient when investigating and fixing test failures, but perhaps it must be tweaked further to show paths if and only if it is appropriate.

Merge with stable

convert: make hg sink cleanup safer

Merge with stable

graphlog: fix output when both a limit and a path are provided Limit was interpreted as absolute, from the topmost revision, without counting the number of revisions matching a given file. Which caused "glog -lN file" to show sometimes less than N csets if the file was not modified in all of the N previous csets. glog will now match the behavior of log.

mercurial.iss: update locale location

test-share: Solaris sed can't handle never-ending lines

test-rollback: Ignore hook exit code Failing hooks returned 255 on Solaris

Merge with crew-stable

test-hook: Don't use extended regular expressions with grep Using egrep fixes failure on Solaris

hgweb: Make get_mtime use repository to find store path. It was calculating it directly, which is redundant and caused it to break with shared repositories.

make ui.interactive() return false in case stdin lacks isatty