merge with stable

dispatch: show deprecation warning if command has no attributes (issue5137) norepo/optionalrepo/inferrepo were removed by aa73d6a5d9ea, which would be significant API change. This patch tries to avoid crash even if ancient third-party extensions are enabled.

dispatch: extract function that tests command attributes This function will host the compatibility layer for old third-party commands. See the next patch for details.

dispatch: make cmdalias forward command attributes to function This delays resolution of command attributes so that missing attributes can be warned only when necessary.

run-tests: use different chg socket directories for different tests Before this patch, if --chg or --with-chg is specified, all tests are using the same chgserver socket. Since the chg client holds a lock when it starts a new server, and every test needs at least a new chg server due to different HGRCPATH affecting the confighash, the result is a lot of tests will be timed out if -j is large (for example, 50 or 100). This patch solves the issue by using different chg socket directories for different tests.

tests: ensure run-tests handles multiple lines of churn 1ad0ddf8cccc added the ability to remember lines (including their flags, like glob) and tolerate them even if the output order varies. This test ensures that the code will continue to work in the future.

subrepo: adapt to git's recent renames-by-default Git turned on renames by default in commit 5404c11 (diff: activate diff.renames by default, 2016-02-25). The change is destined for release in git 2.8.0. The change breaks test-subrepo-git, which test specifically that a moved file is reported as a removal and an addition. Fix by passing --no-renames (available in git since mid 2006) to the diff commands that don't use --quiet (should make no difference for those).

graphmod: move the graphstyle options to experimental I let this slip in the [ui] section during the review, as far as I understand we don't plan to actually support customisation of the output on we are happy with our choice. The option are just here to help people tests various options so we can decide which one we'll actually use. I'm moving the config option in the experimental section to make this clearer and avoid making them part of the public API by mistake.

tests: remove obsolete uses of HGTEST_RUN_TESTS_PURE 342ab95a1f4b added arguments to run-tests.py calls to propagate --pure when --with-hg is not used. 0622d6e134fb unified places to use --with-hg via rt(), which obviated the need for passing HGTEST_RUN_TESTS_PURE.

test-debugextensions: passes with byte-compilation disabled (issue5147)

rpms: add hgext3rd to mercurial.spec Otherwise build process fails with somethings like this: RPM build errors: Installed (but unpackaged) file(s) found: /usr/lib64/python2.7/site-packages/hgext3rd/__init__.py /usr/lib64/python2.7/site-packages/hgext3rd/__init__.pyc /usr/lib64/python2.7/site-packages/hgext3rd/__init__.pyo make: *** [docker-fedora20] Error 1

debugobsolete: add an option to show marker index A bigger picture is the ability to be delete an arbitrary marker form the repo's obsstore. This is a useful debug ability and it needs a way to indentify the marker one wants to delete. Having a marker's index provides such an ability.

tests: explicitly flush output streams Some tests fail while running with chg because they do not flush their output streams. chgserver will make sure ui.flush is called after dispatch, but not after {ui,repo}setup. For other non-ui streams, it should be explicitly flushed since the request handler will use os._exit. This patch adds explicit flushes in test-bundle2-format.t, test-extension.t and test-obsolete.t. It will fix most test cases of them when running with chg.

test-hgrc: do not print ui.plain() flag at uisetup() This makes the test pass with chg. Since uisetup() is run per process, "hg showconfig" does not always call uisetup().

tests: make config/help tests pass even if chgserver extension is loaded It isn't important for these tests if "extensions.chgserver=" is set or not. So let's make them ignore such lines.

setup: alphabetize hgext packages

remove: add progress support

remove: queue warnings until after status messages (issue5140) (API) Before this change, warnings were interspersed with (and easily drowned out by) status messages. API: abstractsubrepo.removefiles has an extra argument warnings, into which callees should append their warnings. Note: Callees should not assume that there will be items in the list, today, I'm lazily including any other subrepos warnings, but that may change. cmdutil.remove has an extra optional argument warnings, into which it will place warnings. If warnings is omitted, warnings will be reported via ui.warn() as before this change (albeit, after any status messages).

tests: include progress for test-remove

chg: allows default hg path to be overridden Before this patch, chg will fall back to "hg" if neither CHGHG nor HG are set. This may have trouble if the "hg" in PATH is not compatible with chg, which can happen, for example, an old hg is installed in a virtualenv. Since it's very hard to do a quick hg version check from chg, after discussion in IRC with smf and marmoute, the quickest solution is to build a package with a hardcoded absolute hg path in chg. This patch makes it possible by adding a C macro HGPATH.

chg: define PRINTF_FORMAT_ for non gnu C compiler Before this patch, if __GNUC__ is not defined, PRINTF_FORMAT_ will not be defined and will cause compilation error. This patch solves the issue by making sure PRINTF_FORMAT_ is defined. It allows chg to be compiled with tcc (http://bellard.org/tcc/) by: tcc -o chg *.c

chgserver: drop bundle.mainreporoot config Before this patch, although chgserver drops repo object by setting it to None, there is another side effect loading a repo: setting bundle.mainreporoot. This patch explicitly sets it to empty to undo the side effect. It will make chg pass test-strip.t.

chgserver: drop old hack to recreate ui on HGPLAIN change It's been superseded by config/env hash. A log message is simplified as we no longer need a diff of environment variables. Also, _loadnewui() is changed to require args since it can't copy old --config values without args.

graphmod: allow edges to end early Rather than draw an edge all the way to the bottom of the graph, make it possible to end an edge to parents that are not part of the graph early on. This results in a far cleaner graph. Any edge type can be set to end early; set the ui.graphstyle.<edgetype> parameter to the empty string to enable this. For example, setting the following configuration: [ui] graphstyle.grandparent = : graphstyle.missing = would result in a graph like this: o changeset: 32:d06dffa21a31 |\ parent: 27:886ed638191b | : parent: 31:621d83e11f67 | : o : changeset: 31:621d83e11f67 |\: parent: 21:d42a756af44d | : parent: 30:6e11cd4b648f | : o : changeset: 30:6e11cd4b648f |\ \ parent: 28:44ecd0b9ae99 | ~ : parent: 29:cd9bb2be7593 | / o : changeset: 28:44ecd0b9ae99 |\ \ parent: 1:6db2ef61d156 | ~ : parent: 26:7f25b6c2f0b9 | / o : changeset: 26:7f25b6c2f0b9 |\ \ parent: 18:1aa84d96232a | | : parent: 25:91da8ed57247 | | : | o : changeset: 25:91da8ed57247 | |\: parent: 21:d42a756af44d | | : parent: 24:a9c19a3d96b7 | | : | o : changeset: 24:a9c19a3d96b7 | |\ \ parent: 0:e6eb3150255d | | ~ : parent: 23:a01cddf0766d | | / | o : changeset: 23:a01cddf0766d | |\ \ parent: 1:6db2ef61d156 | | ~ : parent: 22:e0d9cccacb5d | | / | o : changeset: 22:e0d9cccacb5d |/:/ parent: 18:1aa84d96232a | : parent: 21:d42a756af44d | : | o changeset: 21:d42a756af44d | |\ parent: 19:31ddc2c1573b | | | parent: 20:d30ed6450e32 | | | +---o changeset: 20:d30ed6450e32 | | | parent: 0:e6eb3150255d | | ~ parent: 18:1aa84d96232a | | | o changeset: 19:31ddc2c1573b | |\ parent: 15:1dda3f72782d | ~ ~ parent: 17:44765d7c06e0 | o changeset: 18:1aa84d96232a parent: 1:6db2ef61d156 parent: 15:1dda3f72782d The default configuration leaves all 3 types set to |. This is part of the work towards moving smartlog upstream; currently smartlog injects extra nodes into the graph to indicate grandparent relationships (nodes elided).

graphmod: allow for different styles for different edge types Rather than draw all edges as solid lines, allow for using different styles for different edge types. For example you could use dotted lines for edges that do not connect to a parent, and dashed lines when connecting to a grandparent (implying missing nodes in between). For example, setting the following configuration: [ui] graphstyle.grandparent = : graphstyle.missing = . would result in a graph like this: o changeset: 32:d06dffa21a31 |\ parent: 27:886ed638191b | : parent: 31:621d83e11f67 | : o : changeset: 31:621d83e11f67 |\: parent: 21:d42a756af44d | : parent: 30:6e11cd4b648f | : o : changeset: 30:6e11cd4b648f |\ \ parent: 28:44ecd0b9ae99 | . : parent: 29:cd9bb2be7593 | . : o . : changeset: 28:44ecd0b9ae99 |\ \ \ parent: 1:6db2ef61d156 | . . : parent: 26:7f25b6c2f0b9 | . . : o . . : changeset: 26:7f25b6c2f0b9 |\ \ \ \ parent: 18:1aa84d96232a | | . . : parent: 25:91da8ed57247 | | . . : | o-----+ changeset: 25:91da8ed57247 | | . . : parent: 21:d42a756af44d | | . . : parent: 24:a9c19a3d96b7 | | . . : | o . . : changeset: 24:a9c19a3d96b7 | |\ \ \ \ parent: 0:e6eb3150255d | | . . . : parent: 23:a01cddf0766d | | . . . : | o---+ . : changeset: 23:a01cddf0766d | | . . . : parent: 1:6db2ef61d156 | | . . . : parent: 22:e0d9cccacb5d | | . . . : | o-------+ changeset: 22:e0d9cccacb5d | . . . . : parent: 18:1aa84d96232a |/ / / / / parent: 21:d42a756af44d | . . . : | . . . o changeset: 21:d42a756af44d | . . . |\ parent: 19:31ddc2c1573b | . . . | | parent: 20:d30ed6450e32 | . . . | | +-+-------o changeset: 20:d30ed6450e32 | . . . | parent: 0:e6eb3150255d | . . . | parent: 18:1aa84d96232a | . . . | | . . . o changeset: 19:31ddc2c1573b | . . . .\ parent: 15:1dda3f72782d | . . . . | parent: 17:44765d7c06e0 | . . . . | o---+---+ | changeset: 18:1aa84d96232a . . . . | parent: 1:6db2ef61d156 / / / / / parent: 15:1dda3f72782d . . . . . Edge styles can be altered by setting the following one-character config options:: [ui] graphstyle.parent = | graphstyle.grandparent = : graphstyle.missing = . The default configuration leaves all 3 types set to |, leaving graph styles unaffected. This is part of the work towards moving smartlog upstream; currently smartlog injects extra nodes into the graph to indicate grandparent relationships (nodes elided).

chgserver: use global ui instead of repo ui for dispatch.request.ui Before this patch, chgserver will use repo ui as dispatch.request.ui, while req.ui is designed to be global ui without repo config. Passing repo ui as dispatch.request.ui leads to repo.ui being incorrect, which can lead to unwanted results. For example, if the repo config has [extensions], it could affect which localrepository.featuresetupfuncs get executed and the repo may have an incorrect list of supported requirements. This patch changes _renewui to return both global ui and repo ui. The global ui is passed to req.ui, and the repo ui is used to calculate confighash. It will make chg pass test-largefiles-misc.t and test-requires.t, which are both related to repo requirements.

progress: update comment to reflect implementation Progress has been on by default in core for a while, the comment was written before this happened.

tests: remove lines that enable progress extension progress has not been an extension for a long time, it has been integrated into core, thus these lines have not done anything for a while.

run-tests: add support for automatically bisecting test failures

check-code: also ban strcat We're not using it now, so it's easy to ban.

check-code: prevent use of strcpy

osutil: stop using strcpy strcpy is a security vulnerability masquerading as a utility function. Replace it with memcpy since we know how much to copy anyway.

histedit: add a hint about enabled dropmissing to histedit edit comment Adds a hint to histedit comment reminding user about enabled dropmissing. This will make the enabled dropmissing more visible. The example comment: # Edit history between b592564a803c and b54649a8a63f # # Commits are listed from least to most recent # # You can reorder changesets by reordering the lines # # Commands: # # e, edit = use commit, but stop for amending # m, mess = edit commit message without changing commit content # p, pick = use commit # b, base = checkout changeset and apply further changesets from there # d, drop = remove commit from history # f, fold = use commit, but combine it with the one above # r, roll = like fold, but discard this commit's description # # Deleting a changeset from the list will DISCARD it from the edited history!

hghave: change ssl check to just check ssl module Previously, the "ssl" check effectively looked for PyOpenSSL or Python 2.7.9. After this patch, we simply look for just the "ssl" module. After d962e955da08, there have been no references to PyOpenSSL in the tree (the previous usage of PyOpenSSL was to implement ssl support on old, no longer supported Python versions that didn't have an ssl module (e.g. Python 2.4). So, the check for PyOpenSSL served no purpose. Pythons we support ship with the ssl module. Although it may not be available in all installations. So, we still need the check for whether the ssl module imports, hence the hghave check. The main side-effect of this change is that we now run test-https.t (the only test requiring the "ssl" hghave feature) on Python <2.7.9 when PyOpenSSL is not installed (which is probably most installations) and the ssl module is available. Before, we wouldn't run this test on these older Python versions. I confirmed that test-https.t passes with Python 2.6.9 and 2.7.8 on OS X 10.11.

help: document sharing of revlog header with revision 0 The previous docs were incorrect about there being a discrete header on revlogs.

mpatch: move collect() to module level This helps the code read a little more clearly.

mpatch: un-nest the move() method This helps the code read a little more clearly.

mpatch: move pull() method to top level There was no need for this to be nested.

chgserver: use old ui.system if fout is not stdout or needs to be captured Before this patch, chgui will override the system method, forwarding every process execution to the client so sessions and process groups can work as expected. But the chg client will just use stdout, if ui.fout is not stdout or if the output is set to be captured to safe._buffers, the client will not behave correctly. This can happen especially with code prepending "remote:". For example, bundle2 uses ui.pushbuffer, and sshpeer sets fout to ferr. We may have trouble with interactive commands in the fout set to ferr case but if it really bites us, we can always send file descriptors to the client. This patch adds a check to detect the above situations and fallback to the old ui.system if so. It will make chg happy with test-bundle2-exchange.t, test-phases-exchange.t, test-ssh-bundle1.t and test-ssh.t.

node: use byte literals to construct nullid and wdirid Python 3's hex() insists on operating on bytes. This patch gives it what it wants. '' and b'' in Python 2 are equivalent, so this has no impact on Python 2.

tests: try to import modules with Python 3 All of mercurial.* is now using absolute_import. Most of mercurial.* is able to ast parse with Python 3. The next big hurdle is being able to import modules using Python 3. This patch adds testing of hgext.* and mercurial.* module imports in Python 3. As the new test output shows, most modules can't import under Python 3. However, many of the failures are due to a common problem in a highly imported module (e.g. the bytes vs str issue in node.py).

tests: perform an ast parse with Python 3 Previously, test-check-py3-compat.t parsed Python files with Python 2 and looked for known patterns that are incompatible with Python 3. Now that we have a mechanism for invoking Python 3 interpreters from tests, we can expand check-py3-compat.py and its corresponding .t test to perform an additional AST parse using Python 3. As the test output shows, we identify a number of new parse failures on Python 3. There are some redundant warnings for missing parentheses for the print function. Given the recent influx of patches around fixing these, the redundancy shouldn't last for too long.

run-tests: add --with-python3 to define a Python 3 interpreter Currently, very few parts of Mercurial run under Python 3, notably the test harness. We want to write tests that run Python 3. For example, we want to extend test-check-py3-compat.t to parse and load Python files. However, we have a problem: finding appropriate files requires running `hg files` and this requires Python 2 until `hg` works with Python 3. As a temporary workaround, we add --with-python3 to the test harness to allow us to define the path to a Python 3 interpreter. This interpreter is made available to the test environment via $PYTHON3 so tests can run things with Python 3 while the test harness and `hg` invocations continue to run from Python 2. To round out the feature, a "py3exe" hghave check has been added.

crecord: add docblock to handlekeypressed This information is pretty useful when reading the code.

crecord: fix docblock indentation

crecord: clean up empty lines at ends of docblocks

filemerge: indicate that local/other are p1/p2

sslutil: use preferred formatting for import syntax

largefiles: add some docstrings

largefiles: drop partial support for not having a user cache 971c55ce03b8 introduced support for not having a "global" user cache. In the rare cases where the environment didn't provide the location of the current home directory, the usercachepath function could return None. That functionality has since bitrotten and several code paths did not correctly check for usercachepath returning None: $ HOME= XDG_CACHE_HOME= hg up --config extensions.largefiles= getting changed largefiles abort: unknown largefiles usercache location Dropping the partial support for it is thus not really a backward compatibility breaking change. Thus: consistently fail early if the usercache location is unknown. It is relevant to be able to control where the largefiles are stored and how they propagate, but that should probably be done differently. The dysfunctional code just gets in the way.

largefiles: refactor usercachepath - extract user cache path function It is convenient to have the user cache location explicitly.

shelve: adds restoring newly created branch (issue5048) (BC) Before this patch shelve never preserved branch information, so after applying unshelve branch was the same as it was on working copy no matter in which branch shelve took place. This patch makes bare shelving(with no files specified, without interactive,include and exclude option) remembers information if the working directory was on newly created branch ,in other words working directory was on different branch than its first parent. In this situation unshelving restores branch information to the working directory.

shelve: make non bare shelve not saving branch information in bundle This patch prepares for restoring newly created branch only on bare shelve later because information about new-branch will be preserved only when shelve was bare and working copy branch was different than branch of its parent. In other case information about new-branch will be gone, so unshelve will not recognise that shelve was made on new-branch and it will not restore branch information from the bundle to the working directory.

shelve: preserve newly created branch on non-bare shelve in wctx (BC) Before this patch current branch in working context wasnt preserved after shelve, this patch makes it restore after update.

crecord: rewrite a comment about filtering patches I couldn't parse the previous version. I think the rewrite better expresses the intent of that comment.

run-tests: teach _processoutput to handle multiple lines of churn Instead of treating expected output as happening in a precise order, and assuming that if a line is missing it will never happen, assume that expected output is a prioritized list of likely matching lines. This means that if: foo/bar (glob) baz/bad (glob) changes to: baz/bad foo/bar instead of generating: baz/bad foo/bar For which we've lost both (glob) markers, we will match both lines and generate: baz/bad (glob) foo/bar (glob) This retains any special annotations we have for lines.

run-tests: indent _processoutput to aid readability for next patch The next commit will loop over the expected[pos] list, this change makes that change easier to review.

namespaces: fix name/node confusion There was a lot of copy/paste here.

tests: add test for "log -r wdir() -p" (issue4871) The issue has been fixed by 93bcc73df8d5, "cmdutil.changeset_printer: pass context into showpatch()". This patch adds test to prevent future regression.

tests: include modified/added/removed files in "log -r wdir()" output This patch prepares for the test of "hg log -r wdir() -p".