dirstate: reset _lastnormal and _lastnormaltime on write, invalidate, and clear

discovery: avoid discovery when local graph is a subset of remote Immediately sends local's heads to the server to check whether the server knows them all. If it does, we can call getbundle immediately. Interesting test output changes are: - added 1 changesets with 0 changes to 1 files (+1 heads) + added 1 changesets with 0 changes to 0 files (+1 heads) -> The new getbundle() actually fixes a bug vs. changegroupsubset() in that it no longer returns unnecessary files when file revs are reused. warning: repository is unrelated + requesting all changes -> The new use of common instead of bases correctly indicates that an unrelated pull gets all changes from the server.

wireproto: add getbundle() function getbundle(common, heads) -> bundle Returns the changegroup for all ancestors of heads which are not ancestors of common. For both sets, the heads are included in the set. Intended to eventually supercede changegroupsubset and changegroup. Uses heads of common region to exclude unwanted changesets instead of bases of desired region, which is more useful and easier to implement. Designed to be extensible with new optional arguments (which will have to be guarded by corresponding capabilities).

merge with stable

i18n-pt_BR: synchronized with 6783f47d90dd

merge with i18n

i18n-it: synchronized with adf3c4401c5d

osutil: fix up check-code issues

dirstate: flush _lastnormal when we see newer filesystem times

util: add Mac-specific check whether we're in a GUI session (issue2553) The previous test assumed that 'os.name' was "mac" on Mac OS X. This is not the case; 'mac' was classic Mac OS, whereas Mac OS X has 'os.name' be 'posix'. Please note that this change will break Mercurial on hypothetical non-Mac OS X deployments of Darwin. Credit to Brodie Rao for thinking of CGSessionCopyCurrentDictionary() and Kevin Bullock for testing.

rebase: allow for rebasing descendants onto ancestors on different named branches So far we've been denying rebasing descendants onto ancestors, but there are situations in which this kind of operation makes perfect sense to me. Let's say we have made a commit (or more), that belongs to branch 'dev', on top of the named branch 'stable': ... a (stable) - b (dev) but then we realize that b should belong to branch 'stable'. In these cases a rebase means: "move these csets from named branch A to named branch B" and there isn't a valid reason to deny it. This patch basically doesn't block it, if source and destination are on different named branches. The old behaviour still applies for rebases across the same named branch. Can you think of any tricky corner cases in which this new behaviour could lead to problems? (I bet there are tons of them...) By the way, I created a brand new .t because I feel there should be more tests I can't think of at the moment.

bdiff.c: rename all variables which hold a hash value to "hash"

bdiff.c: use unsigned arithmetic for hash computation Signed integer overflow is undefined in C.

bdiff.c: cast to unsigned char when computing hash value

bdiff.c: make all local functions static

merge with stable

debugbundle: fix up long line caught by check-code

wireproto: avoid naked excepts

mq: fix typo in docstring

commands: add debugbundle command Lists ids contained in a bundle file. Useful for testing bundle-related commands.

wireproto: add known([id]) function known([Node]) -> [1/0] Returns 1/0 for each node, indicating whether it's known by the server. Needed for new discovery protocols introduced in later patches.

wireproto: fix decodelist to properly return empty list Needed by tests for next patch introducing known().

wireproto: fix handling of '*' args for HTTP and SSH

debug: add debugwireargs to test argument passing over the wire Tests argument passing locally, via HTTP, and via SSH. This is mainly preparation for the next patch.

sshserver: drop unnecessary line

merge: avoid unlinking destination of merge when case changes (issue2715)

dirstate: introduce a public case normalizing method

changegroup: minor cleanups

repair: revlog has an iterator

changegroup: fix leftover from delta read

changegroup: drop expensive redundant usage of readdelta for progress

changegroupsubset: use manifest.readfast to simplify collector

manifest: add readfast method

changegroupsubset: simplify filenode_collector

changegroupsubset: more renaming

changegroupsubset: simplify prune Ancestors of nodes linked to commonrevs can be expected to be linked to commonrevs. Walking graphs of each revlog looking for rare/nonexistent outliers is overkill.

changegroupsubset: more minor cleanups - remove more excessive comments - simplify some sorting operations - rename some variables - replace identity with a lambda

changegroupsubset: minor cleanups - move some variable declarations - drop some excessive comments - use standard variable naming

strip: simplify collectone

dirstate: avoid a race with multiple commits in the same process (issue2264, issue2516) The race happens when two commits in a row change the same file without changing its size, *if* those two commits happen in the same second in the same process while holding the same repo lock. For example: commit 1: M a M b commit 2: # same process, same second, same repo lock M b # modify b without changing its size M c This first manifested in transplant, which is the most common way to do multiple commits in the same process. But it can manifest in any script or extension that does multiple commits under the same repo lock. (Thus, the test script tests both transplant and a custom script.) The problem was that dirstate.status() failed to notice the change to b when localrepo is about to do the second commit, meaning that change gets left in the working directory. In the context of transplant, that means either a crash ("RuntimeError: nothing committed after transplant") or a silently inaccurate transplant, depending on whether any other files were modified by the second transplanted changeset. The fix is to make status() work a little harder when we have previously marked files as clean (state 'normal') in the same process. Specifically, dirstate.normal() adds files to self._lastnormal, and other state-changing methods remove them. Then dirstate.status() puts any files in self._lastnormal into state 'lookup', which will make localrepository.status() read file contents to see if it has really changed. So we pay a small performance penalty for the second (and subsequent) commits in the same process, without affecting the common case. Anything that does lots of status updates and checks in the same process could suffer a performance hit. Incidentally, there is a simpler fix: call dirstate.normallookup() on every file updated by commit() at the end of the commit. The trouble with that solution is that it imposes a performance penalty on the common case: it means the next status-dependent hg command after every "hg commit" will be a little bit slower. The patch here is more complex, but only affects performance for the uncommon case.

changegroupsubset: extranodes are no longer needed

strip: remove usage of extranodes Instead of computing the exact set of missing revlog revisions, we only compute the set of missing/broken changesets. The resulting bundle can be slightly bigger but we will be able to get rid of the ugly extranodes handling in changegroupsubset.

patch: move closefile() into patchfile.close()

patch: inline patchfile.hashlines()

patch: fix hunk newlines when parsing hunks, not in iterhunks()

i18n: register new template keywords for translation

annotate: catch nonexistent files using match.bad callback (issue1590)

merge with stable

wix: add new file templates/paper/bookmarks.tmpl This file was snuck onto the stable branch with 38c9837b1f75

merge with stable

push/outgoing: print remote target path even if there's an error (issue2561) This is a simple patch to make hg push/hg outgoing print their remote target path even if the operation fails. I'm not sure if the original behavior was by design. This patch also changes one test to reflect the changed behaviour.

subrepo: recognize scp-style paths as git URLs

convert: add svnrev, svnpath and svnuuid template keywords $ hg -R A-hg log --template '{rev} {svnuuid}{svnpath}@{svnrev}\n' 10 644ede6c-2b81-4367-9dc8-d786514f2cde/trunk@10

convert/svn: extract revsplit() in a function

transplant: add "transplanted" keyword $ hg log --template '{rev} {transplanted}\n' 7 a53251cdf717679d1907b289f991534be05c997a

merge with i18n

i18n: merge with stable

i18n-da: synchronize with e9628665b670

i18n: don't mark trivial string for translation

i18n: merge with mpm

i18n-de: correct translation for some fuzzy strings

i18n-de: synchronize with 994510694b1d

i18n-de: added translations concerning hardlinks

i18n-de: added more translations concerning rebase

i18n-de: added translations concerning rebase

i18n-de: added translations concerning email sending

i18n-de: added translations concerning patches

i18n-de: added translations concerning win32text and .hgeol

i18n-de: added translations for Perforce conversion options

i18n-de: added translations for Subversion conversion options

i18n-de: added translation for CVS conversion options

i18n-de: added translations for convert.hg.* options

i18n-de: translate some strings

revset: fix a number of highly dubious continue statements This patch definitely needs a review and would also benefit from some new testsuite entries.

graphlog: do not swallow all TypeError exceptions (fix 29c800ee54cf)

archive: use hardcoded constants when creating .zip archives Do not rely on local stat constants, which may differ.

hgweb: support multiple directories for the same path [paths] /dir = /path/1/*, /path/2/*

templates: provide granularity for future values for age filter

revset: report a parse error if a revset is not parsed completely (issue2654)

config: use util.posixfile

bundle: update current bookmark to most recent revision on current branch We check if the current bookmark is set to the first parent of the dirstate. Is this the case we move the bookmark to most recent revision on the current branch (where hg update will update you to).

localrepo: do not update bookmarks in addchangegroup We want to update the current bookmark to the most recent revision on current branch unless there is a remote bookmark that points to a different descendant. Addchangegroup is called before we can check for remote bookmarks. We don't update the bookmark in addchangegroup anymore to allow proper updating of bookmarks in pull.

rebase: add -m/--message to rebase --collapse (issue2389) When collapsing changesets with rebase, you get a chance to edit the commit message manually, but there is no way to pass this message from the command line. This patch adds a `--message` (with short form `-m`) and `--logfile` (with short form `-m`) options to the rebase command. These options suppresses the generation of the default commit message, and instead use the message provided in the option (in case of `-m`) or in the file it points to (in case of `-l`). If you use this option without the `--collapse` option, it will raise an error. Options documentation edited by Patrick Mezard <pmezard@gmail.com>

Updating hgext.extdiff to use revsets

merge with stable

relink: mark abort message for translation

relink: avoid trying to lock the same repo twice

relink: format reclaimed byte count nicely

test-relink: remove unnecessary ui.username setting

test-https: match output from 31eac42d9123

merge with stable

merge with crew

convert/svn: fix _iterfiles() output in root dir case (issue2647) When converting directory additions/replacement with project directory set to root, _iterfiles() sometimes returned paths starting with a slash making following svn calls to fail. I could not reproduce the issue with hand-crafted repositories. Report and first analysis by Clinton Chau <clinton@clearcanvas.ca>

eol: no need to accumulate files when checking all changesets Just check the files touched by the changesets.

eol: improve hook failure output Before, only the first failure was reported: abort: b.txt should not have CRLF line endings while now all of them are listed: abort: end-of-line check failed: d.txt in a7040e68714f should not have CRLF line endings b.txt in fbcf9b1025f5 should not have CRLF line endings As first suggested by Antoine Pitrou <solipsis@pitrou.net>

merge with stable

bookmarks: remove API limitation in setcurrent setcurrent refuses to set a new current bookmark if the current bookmark points to the current dirstate. This restriction is not needed. A current bookmark can point to a different bookmark.

bookmarks: separate bookmarks update code from localrepo's pull. We explicitly want to update bookmarks from a remote. This will avoid duplicate calls to listkeys if we clone (which calls pull) and keep bookmark related code together.

tests: add tests for bookmarks support in hg identify

identify: list bookmarks for remote repositories

hgignore: ignore dll files generated under cygwin

pager: don't page stderr if it's being redirected (issue2541)

color: port to using ctypes (issue2687) replacing usage of pywin32, which was removed for Mercurial 1.8

hgweb: drop use of super() for Exception base class In Python 2.4, Exception is an old-style class and doesn't work with super.

diff: make diff -c aware of revision sets

color: wrap lines in docstring

merge with stable

setup: fix mac build broken by e42d18538e1d Sometimes xcodebuild prints warnings to stderr, but runcmd() assumes anything printed to stderr implies failure. Since runcmd() was originally only intended to run hg, this was fine until it was pressed into service for running xcodebuild. Thus: split runcmd() into two parts: runcmd(), which does the minimal amount of work to run a subprocess, and runhg(), which calls runcmd().

color: don't mention internal function in docstring The render_text function actually never existed, the docstring probably meant the render_effects function instead.

keyword: offer additional datefilters when the extension is enabled Before the additional datefilters (utcdate, svnisodate, svnutcdate) were used when kwtemplater was initialized. Now they always be used once the extension is enabled.

keyword: docstrings for additional date filters

mq: do not let qrefresh write bad patch

graphlog: give meaningful error if more than one file is given

merge with stable

test-https: update to match output from 3f6a4579f803

tests: respect new hg clone listkey queries in http-proxy test

bookmarks: discard current bookmark if absent from the bookmarks (issue2692) After a rollback, the current bookmark might be absent from the bookmarks file. In that case we discard it instead of displaying a traceback during commit.

fix test-highlight after bookmark support change for hgweb

merge with stable

eol: do not abort when win32text is found, only warn The win32text extension does not break eol or vice-versa, so it is not a fatal error to have both of them enabled. It's just folly. So spewing warnings in this condition is preferrable to aborting. When both extensions are enabled, the user now sees: % hg st the eol extension is incompatible with the win32text extension win32text is deprecated: http://mercurial.selenic.com/wiki/Win32TextExtension M hgext/eol.py