zeroconf: better fix for readName error Stop parsing when an unknown type is seen, since it is impossible to resynchronize on the stream.

zeroconf: try utf-16 decoding if utf-8 fails

import: import each patch in a file or stream as a separate change Supports hg export <revrange>, mail messages, and mailboxes. Does not support multiple patches in a single MIME attachment. Closes issue167.

ui: look before you leap on sys.stdout.closed

mq: fix deprecation mention, line length

merge with mpm

addbranchrevs: fallback for older servers

add -b/--branch option to clone, bundle, incoming, outgoing, pull, push

subrepo: change default path in hgrc of subrepo after cloning Previous behavior was to put in the cloned subrepos the path found in the original main repo. However it isn't valid for relative path and it seems more logical to reference instead the subrepos working copy path of the original main repo.

streamclone: allow uncompressed clones by default

help: Improved body text

commands: Try to improve help options text for basic commands Mostly rewrite -r text

check-code.py: make help strings consistent

check-code.py: escape backslash

qnew: ignore force option This makes the default behavior the same as qnew --force, and deprecates the force option.

strip apostrophes from an error messages

mq: fix typo, thanks pmezard

commands: annotate follows by default, separate -f/--file option

Merge with crew-stable

test-mq-qrefresh: test qrefresh when tip != qtip

mq: remove qrefresh slow path (issue2025) The slow path/fast path distinction has existed since mq was added to Mercurial in 2006. The slow path was used whenever the refreshed revision wasn't the tip

interpret repo#name url syntax as branch instead of revision Previously, the name part of an repo#name url was interpreted as a revision, similar to using the --rev option. Now it is instead looked up as a branch first, and if that succeeds all the heads of the branch will be processed instead of just its tip-most head. If the branch lookup fails, it will be assumed to be an revision as before (e.g. for tags).

dispatch: provide help for disabled extensions and commands Before a command is declared unknown, each extension in hgext is searched, starting with hgext.<cmdname>. If there's a matching command, a help message suggests the appropriate extension and how to enable it. Every extension could potentially be imported, but for cases like rebase, relink, etc. only one extension is imported. For the case of "hg help disabledext", if the extension is in hgext, the extension description is read and a similar help suggestion is printed. No extension import occurs.

extensions: refactor disabled()

mq: don't warn on qpush against a branch head

mq: deprecate qinit and qcommit

mq: deprecate qsave, qrestore and related options

mq: add --mq option to some commands This causes them to operate on the queue repository as qcommit does, and is nicer than hg -R $(hg root)

pull: with -u and -r, update to the first revision given This makes pull consistent with clone. The actual change is in mercurial.hg.parseurl, changing its checkout result, which is only used for pull.

localrepo: inline single-use nested function

localrepo: unify changegroup and changegroupsubset code paths a bit

merge: add hints about the use of 'hg heads' to find the rev to merge

prepush: warn about every new outgoing named branch, not just the first

prepush: rename variables, refactor

rebase: add --detach option to detach intermediate revisions (issue1950) When rebasing an intermediate revision, rebase keeps a parent relationship with the original parent. This option forces the removal of this relationship. In more depth, it 'fakes' null merges between the target revision and the ancestors of source, dropping every change from the ancestors. The result is that every change in source and its descendants will be rebased, ignoring the changes in its ancestors.

rebase: refactoring Separate rebase-specific functions, in order to provide a better base to extend rebase's capabilities. Note that this will be useful especially to share functions with 'adapt'.

commands: do all branch heads by default, demote topological to -t/--topo

commands: actually implement --closed for topological heads

commands: externalize branchheads so we can do it for all branches at once

commands: move inactive/closed out so they can have more effect

commands: don't do too much work for error messages

commands: allow --closed even for topological heads displays But it actually only starts working later on.

util: make spawndetached() handle subprocess early terminations The file-based synchronization introduced by e22695b4472f hangs when the child process fails before terminating the handshake, which the previous pipe-based version handled correctly. To fix this, the parent polling loop was fixed to detect premature terminations of the child process.

zeroconf: do not try to encode encoded strings (issue1942)

zeroconf: override default destination folder on clone

merge with main

merge with main

fix spaces/identation issues

Merge with crew

Merge with crew

run-tests: kill daemons on ^C with -j. This is not perfect, but the problem is fundamentally racy.

commands: fix more changeset header information in 'hg help export'

commands: fix the list of changeset header information in 'hg help export'

graphlog: remove unused import

keyword: fix utcdate filter So far the utcdate filter only omitted the timezone, now it actually converts to utc. Thanks to Kevin Grover for pointing out the issue.

commands: always order heads recent to oldest

commands: fix up stupidly untested heads simplification

revlog: fix up previously stupid API change

commands: simplify heads a little bit before I start hacking it up

localrepo: add a quick docstring for localrepo.branchmap()

commands: deprecate the heads -a option Hopefully we can remove it entirely in two releases or so.

revlog: add a fast path for checking ancestry

subrepo: correct copyright

record: function variable naming & signature cleanup. No functionality change. All tests pass.

add options dict to localrepo.store.opener and use it for defversion

merge with main

localrepo: minor formatting - remove double space

coverage.py: fixed syntax error

merge with stable

zeroconf: gethostbyaddr may also fail with socket.herror

Merge with stable

commands: label from user is in local encoding repo.branchtags().keys() are in UTF-8, so the label should be converted to UTF-8 before checking for a naming conflict. Keep the original label for ui.status()

merge with main

copies: check if revisions are related (bug found with pylint)

Merge with stable

Added signature for changeset 4aa619c4c2c0

Added tag 1.4.3 for changeset 4aa619c4c2c0

Merge with i18n

Merge with i18n

merge with i18n stable

i18n-pt_BR: synchronized with b08ffd27dfc8

Merge with stable.

i18n: fix typo in German translation

i18n-ja: synchronized with e7727a545c48

Merge with stable

templates: rename `Last change' column in hgwebdir repository list. This patch changes column headers in the templates that previously said `Last change' to `Last modified'. Neither code nor functionality are changed other than that. For some time now, I have been annoyed by the fact the `Last change' column didn't list the age of the youngest changeset in the repository, or at least tip. It just occurred to me that this is because the wording is slightly misleading; what the column in fact lists is when the repository was last *modified*, that is, when changesets was last added or removed from it. The word `change' can be understood as referring to the changeset itself. Using `changed' would be ever so slightly less amigous. However, the standard nomenclature in this case is `modification date' and `Last modified', which is incidentally entirely unambigous. Hence, `Last modified' is the wording used.

run-tests: split tests/blacklist in tests/blacklists/* Following discussions with Gilles Morris [1], it seems that it is preferable to use several blacklist files in a blacklists/ directory. It is easier to add an unversioned file for experiments than modifying a tracked file. Also fall back to a simpler syntax, giving up ConfigParser, now that section names are not needed anymore. And allow --blacklist parameter to be a complete path, instead of only one of the filenames contained in tests/blacklists/ [1] http://www.selenic.com/pipermail/mercurial-devel/2009-December/017317.html

subrepo: fix errors reported by pylint

copies: revert d1cd09bb4d18 The behavior of rebase both before and after the fix was actually incorrect, and this change regressed other parts of copy behavior.

Merge with i18n

Merge with stable

config: handle short continuations (issue1999) Thanks to Greg Ward for spotting and testing

copies: fix issue1994 Unscramble divergence test

i18n-sv: Translated twelve basic commands The help text and output strings surrounding these commands are now available in glorious Swedish: outgoing, paths, recover, rename, resolve, revert, rollback, root, showconfig, tag, tags, tip This also syncs with 279c8a73fde1.

bugzilla: lowercase mysqldb var

check-code: del isn't a function

check-code: fix copyright date

test-minirst: hide trailing whitespace from check-code

zeroconf: flag third-party Zeroconf.py for ignore in check-code

check-code: add some ignore hints

check-code: check thyself

Merge with stable.

patchbomb: document [patchbomb] config section for addresses

hgrc.5: whether bcc is prompted for is up to the extension There is no code in mail.py that forbids an extension to prompt for bcc.

many, many trivial check-code fixups

Introduce check-code.py check-code is a simple regex-based framework for checking our code and tests for common style and portability errors. Currently, it knows a fair amount about our Python and C style, and a little about common shell script portability problems.

merge with stable

test-encode: Use tr chacter classes instead of character ranges Solaris ...

Merge with stable

templates: do not add "ago" twice After 1d75c683ada1, "ago" is added by the ago template filter.

Merge with stable

tests: Don't use $PWD $PWD doesn't work with solaris sh - use `pwd` instead

mq: Do not translate import message that are appended to commits. We should not translate the "imported patch" message. The translated message confuses the detection whether the user has not updated the commit message yet. We try to avoid to translate generated commit messages.

subrepo: handle svn externals and meta changes (issue1982) - Detect changes to meta on regular and external entries - Do not try to commit external entries

subrepo: svn xml output is much easier to parse That's especially true with status flags: there are different fields and values for regular changes, meta changes, externals changes and externals meta changes.

subrepo: make svn use C locale for portability My reasons to do c2e2a5e6c36b were wrong. We want to have non-localized svn output while preserving current locale for input arguments and svn internals. This is achieved by setting LC_MESSAGE=C.

Merge with stable

summary: L10N messages hide clean-ness of workdir from 'hg summary' 'hg summary' recognizes workdir as clean by searching for 'clean' is in own output text. But 'clean' is one of I18N target texts, so 'hg summary --quiet' behaves incorrectly in non-English locale.

mercurial.el: add forgotten version number The number was accidentally removed in the change from GPLv2 to GPLv2+ in 25e572394f5c.

test-subrepo-svn: remove bashism

Merge with crew-stable

test-patch-offset: Don't use unified diff Solaris diff -u likes to be "helpful" and tell that "No differences encountered"

Merge with stable

Update license to GPLv2+

copies: speed up copy detection On some large repos, copy detection could spend > 10min using fctx.ancestor() to determine if file revisions were actually related. Because ancestor must traverse history to the root to determine the GCA, it was doing a lot more work than necessary. With this replacement, same status -r a:b takes ~3 seconds.

hgweb: request: strip() form values Entering "<correct_cset_hash> " in the search form was not returning anything. This happens relatively often, due to HTML formatting: when copy/pasting a cset hash from the web, selection might contain surrounding spaces.

templatekw: fix extras, manifest and showlist args (issue1989) Removing the explicit parameters from keywords signature is easier than copying and reproducing the original argument list.

merge with i18n stable

i18n-pt_BR: synchronized with 279c8a73fde1

Merge with stable

share: be more explicit about the dangers of rollback.

mercurial.spec: Alignment with AIX spec by Jim Hague * Require Python 2.4 or later * Note that hgk uses tk - but don't require it * Remove bashism to support ksh * Install mq.el * Handle uncompressed man pages

hgweb: changenav: separate pages before and after the current position This should allow styles to customize more precisely navigation bar, for example inserting a cursor to show where we are in the navigation bar.

Work around AIX shell builtin printf not handling \NNN. On AIX, ksh builtin printf does not understand \NNN. Some tests use this to generate test data, and so fail on AIX. Rework these tests to use python to generate the correct characters. This fixes the tests on AIX and should be more generally portable.

merge with stable

subrepo: fix merging of already merged subrepos (issue1986) This fixes a bug seen when merging a main repo which contains a subrepo when both repos have been merged before. Each repo (main and sub) has two branches, both of which have been merged before. In a subrepo, if the revision to merge to is an ancestor of the current rev, then the merge should be a noop. Test provided by Steve Losh.

Merge with crew-stab le

cmdutil: expand style paths (issue1948) Fix suggested by Andrea Riciputi <andrea.riciputi@gmail.com>

hgweb: add a test for search logs

hgweb: add less/more links for search logs (issue1972)

hgweb: add less/more links to shortlog/filelog nav

hgweb: make graph page size equal to shortlog

merge with stable

ui: just return it if it's already a bool

demandimport: ignore _winreg (used in python-2.7 mimetypes)

shrink-revlog: add --dry-run option

cmdutil: hide child window created by win32 spawndetached() Hiding the child process window is not strictly necessary but it avoids opening an empty shell window when running hg serve as well as a task in the task bar. The window is hidden after the process is already started causing a single flicker.

Find right hg command for detached process On Windows, Mercurial can be run from the python script of from a frozen executable. In the first case, we have to call the python interpreter since the script is not executable. Frozen executable can be called directly. Fix 3/3 for issue421

cmdutil: replace unix pipe handshake with file lock The pipe handshake cannot be implemented with Windows detached process, which disables handle inheritance. Fix 2/3 for issue421

serve: add and use portable spawnvp replacement There is no standard python command to really detach a process under Windows. Instead we use the low level API wrapped by subprocess module with all necessary options to avoid any kind of context inheritance. Unfortunately, this version still opens a new window for the child process. The following have been tried: - os.spawnv(os.P_NOWAIT): works but the child process is killed when parent console terminates. - os.spawnv(os.P_DETACH): works on python25, hang on python26 when writing to the hgweb output socket. - subprocess.CreateProcess() hack without shell mode: similar to os.spawnv(os.P_DETACH). Fix 1/3 for issue421

shrink-revlog: help/doc tweaks - make it look right in 'hg help extensions' - clarify --revlog option - add a main program to tell users this is no longer a standalone script

Add shrink-revlog.py to win32 installer

shrink-revlog: make it work on windows (issue1976)

merge main with stable

merge crew together

merge stable together

merge with stable

shrink-revlog: avoid cryptic error when used as an extension When shrink-revlog is used as an extension, it would produce an error when switching back to stable because the script will try to execute itself.

transaction: initialize self.journal to None after deletion This avoid a warning when having an empty transaction, where hg would try to delete the journal twice.

subrepo: fix typo in help text

Merge with stable

relink: ensure consistent candidate order

commands: fix typo in help text

color: colorize output of hg resolve -l

color: improve coding style

Merge with crew-stable

ui: display progress with decimal notation

win32: close file when leaving _getfileinfo()

Add support for relinking on Windows. Test and minor code change by Patrick Mézard <pmezard@gmail.com>

Add test for relink extension

contrib: small documentation fixes in shrink-revlog.py

contrib: turn shrink-revlog.py into an extension

contrib: fix error handling in shrink-revlog.py to be more hg-like

contrib: use ui to write in shrink-revlog.py

Merge with stable.

Update test and man page for multiple matching encode/decode filters

commands: fix help string for pull -u and unbundle -u. They update to new branch head, not new tip -- same as 'hg update'.

Update copyright year to 2010

httprepo: always store the response url (issue1968) Fixes a regression caused by 54b518fc6671

Merge with stable

commands: update copyright year to 2010

test-convert-svn-branches: filter unwanted output Windows cmd.exe emits a different message than sh: """ foobar is not recognized as an internal or external command, operable program or batch file. """ Filter it.

Merge with crew-stable

patch: explicitely close input patch files when leaving If applydiff() raises an exception, the opened patch file is kept alive in the exception context. If it is a temporary file (for instance supplied by import command with stdin input), Windows cannot clean it up.

Merge with crew-stable

hgweb: improve IIS PATH_INFO fix ee8af8a4d905 (issue1580) Copy PHP IIS detection.

Minor svn conversion tests cleanup

subrepo: force en_US.UTF-8 locale when calling svn Parser only knows about en_US output. Forcing the encoding to UTF-8 might not be the best thing to do since the caller may receive some of the subversion output, but at least it should prevent conversion errors from svn client.

test-subrepo-svn: normalize OS specific path separator

subrepo: normalize svn output line-endings

test-subrepo-svn: avoid pushd/popd

Merge with stable

Added signature for changeset 296a0b14a686

Added tag 1.4.2 for changeset 296a0b14a686

Merge with crew

mq: qdiff with the same diff options than qrefresh (issue1350) Specifically, if current patch is a git patch and mq.git=keep, diff in git mode.

mq: upgrade to git patch when necessary (issue767)

patch: support diff data loss detection and upgrade In worst case, generating diff in upgrade mode can be two times more expensive than generating it in git mode directly: we may have to regenerate the whole diff again whenever a git feature is detected. Also, the first diff attempt is completely buffered instead of being streamed. That said, even without having profiled it yet, I am convinced we can fast-path the upgrade mode if necessary were it to be used in regular diff commands, and not only in mq where avoiding data loss is worth the price.

Merge with crew-stable

Merge with stable

mq: preserve --git flag when folding patches Without this, folding a git patch into a regular one downgrades the resulting patch to a regular patch.

mq: preserve --git flag when merging patches Without this, merging a patch queue without diff.git=1 downgrades all git patches to regular patches, losing data in the process.

mq: stop caching and sharing diff options Several mq commands mutate the shared options which causes hard to track bugs when multiple calls are combined.

subrepo: mark string for translation and fix case

Add RelaxNG schema to win32 installer

Merge with mpm

Merge with crew-stable

copies: don't report copies with unrelated branch

subrepo: Subversion support

subrepo: add table-based dispatch for subrepo types

dirstate: don't check state of subrepo directories

subrepo: load from a context where the subrepo exists

subrepo: fix includes support in .hgsub

Merge with stable

i18n: update po files for 703db37d186b and 0ddbc0299742

Merge with i18n

Merge with stable

Merge with i18n.

Merge with i18n-stable

i18n: merge with stable

i18n-da: import strings from default branch

i18n-sv: synchronized with 37679dbf2ee3

merge with i18n stable

i18n-it: removed extraneous '\r' from message

i18n-pt_BR: synchronized with 37679dbf2ee3

Added RelaxNG schema for hg log XML output format

Bugfix and test for hg log XML output

Added output of "extras" to hg log XML output

Merge with stable

run-tests.py: clears http_proxy for all tests

windows: expand environment vars in find_exe This allows us to use ${ProgramFiles} in our [merge-tools] sections so we can provide locale safe program paths.

Merge with stable

run-tests.py: clears http_proxy for all tests

Added XML output for hg log The XML format is based on the one used by Subversion. Currently the closing "</log>" tag is not written, since cmdutil.py does not support the "footer" template variable.

Added support for templatevar "footer" to cmdutil.py

patch: make git diffline generation more explicit

mq: untangle git patch detection

test-mq: better tests delimitation

Merge with crew-stable

test-bisect: test handling of invalid commands without relying on sh