copies: drop the findlimit logic We don't use the limit anymore so we should stop computing that limit. I did not bother measuring the potential performance gain. I am assuming that not running any code will be faster that doing some computation and not using the result.

pathcopies: give up any optimization based on `introrev` Between 8a0136f69027 and d98fb3f42f33, we sped up the search for the introduction revision during path copies. However, further checking show that finding the introduction revision is still expensive and that we are better off without it. So we simply drop it and only rely on the linkrev optimisation. I ran `perfpathcopies` on 6989 pair of revision in the pypy repository (`hg perfhelper-pathcopies`. The result is massively in favor of dropping this condition. The result of the copy tracing are unchanged. Attempt to use a smaller changes preserving linkrev usage were unsuccessful, it can return wrong result. The following changesets broke test-mv-cp-st-diff.t - if not f.isintroducedafter(limit): + if limit >= 0 and f.linkrev() < limit: return None Here are various numbers (before this changeset/after this changesets) source destination before after saved-time ratio worth cases e66f24650daf 695dfb0f493b 1.062843 1.246369 -0.183526 1.172675 c979853a3b6a 8d60fe293e79 1.036985 1.196414 -0.159429 1.153743 22349fa2fc33 fbb1c9fd86c0 0.879926 1.038682 -0.158756 1.180420 682b98f3e672 a4878080a536 0.909952 1.063801 -0.153849 1.169074 5adabc9b9848 920958a93997 0.993622 1.147452 -0.153830 1.154817 worse 1% dbfbfcf077e9 aea8f2fd3593 1.016595 1.082999 -0.066404 1.065320 worse 5% c95f1ced15f2 7d29d5e39734 0.453694 0.471156 -0.017462 1.038488 worse 10% 3e144ed1d5b7 2aef0e942480 0.035140 0.037535 -0.002395 1.068156 worse 25% 321fc60db035 801748ba582a 0.009267 0.009325 -0.000058 1.006259 median 2088ce763fc2 e6991321d78b 0.000665 0.000651 0.000014 0.978947 best 25% 915631a97de6 385b31354be6 0.040743 0.040363 0.000380 0.990673 best 10% ad495c36a765 19c10384d3e7 0.431658 0.411490 0.020168 0.953278 best 5% d13ae7d283ae 813c99f810ac 1.141404 1.075346 0.066058 0.942126 best 1% 81593cb4a496 99ae11866969 1.833297 0.063823 1.769474 0.034813 best cases c3b14617fbd7 743a0fcaa4eb 1101.811740 2.735970 1099.075770 0.002483 c3b14617fbd7 9ba6ab77fd29 1116.753953 2.800729 1113.953224 0.002508 058b99d6e81f 57e249b7a3ea 1246.128485 3.042762 1243.085723 0.002442 9a8c361aab49 0354a250d371 1253.111894 3.085796 1250.026098 0.002463 442dbbc53c68 3ec1002a818c 1261.786294 3.138607 1258.647687 0.002487 As one can see, the average case is not really impacted. However, the worth case we get after this changeset are much better than the one we had before it. We have 30 pairs where improvements are above 10 minutes. This reflect in the combined time for all pairs before: 26256s after: 1300s (-95%) If we remove these pathological 30 cases, we still see a significant improvements: before: 1631s after: 1245s (-24%)

perf: introduce a `--contains` flag to the `perfdirstate` command The new flag benchmark a large amount of `filepath in dirstate` call. This will be useful to compare the Python and Rust implementation of the dirstatemap.

perf: introduce a `--iteration` to `perfdirstate` This flag benchmark an iteration over all the file in the dirstate. This will be useful to compare the Python and the Rust implementation of the dirstate.

tests: open file in binary mode in test-upgrade-repo.t Otherwise the file sizes printed in the output diverge from other platforms. Differential Revision: https://phab.mercurial-scm.org/D7081

hghave: use a native string to invoke the `black` command Windows can't create a process based on bytes. Apparently only the output test in `matchoutput` wants bytes. Differential Revision: https://phab.mercurial-scm.org/D7080

py3: convert cwd to native string when running `fix` This prevents a bunch of failures on Windows. Differential Revision: https://phab.mercurial-scm.org/D7079

format: format commands.py, which recently regressed # skip-blame just reformatting Differential Revision: https://phab.mercurial-scm.org/D7064

py3: fix test-bad-extension expectations for py35 ModuleNotFoundError was introduced in 3.6, so the tests need to expect to see ImportError with 3.5. Differential Revision: https://phab.mercurial-scm.org/D7063

phabricator: add addremoved and addmodified functions These are relatively simple so add them together. As the name indicates, these add phabchange objects for removed and modified files to the phabdiff. Differential Revision: https://phab.mercurial-scm.org/D7049

phabricator: add makebinary and addoldbinary functions These populate the phabchange with the data for a binary file, much as maketext does for text files. Differential Revision: https://phab.mercurial-scm.org/D7048

phabricator: add the uploadfile function This is needed to be able to submit binary files such as images in commits (and also non-UTF-8 text files). One of the primary disadvantages of the current use of createrawdiff is that Phabricator simply drops any binary diffs included in the patch, but not the files, so if such a diff is then committed it corrupts any binary files in it. Differential Revision: https://phab.mercurial-scm.org/D7047

phabricator: add the uploadchunks function Differential Revision: https://phab.mercurial-scm.org/D7046

phabricator: add the maketext function This add the diff data for a text file to a phabchange. Differential Revision: https://phab.mercurial-scm.org/D7045

phabricator: add the phabdiff data structure This holds all the data about a commit, and is passed to the differential.creatediff API. Differential Revision: https://phab.mercurial-scm.org/D7044

phabricator: add the phabchange data structure These store data about individual files in a commit. Differential Revision: https://phab.mercurial-scm.org/D7043

phabricator: add the phabhunk data structure These store the actual diff data (for UTF-8 text files anyway) and are equivalent to hunks in a patch file. Differential Revision: https://phab.mercurial-scm.org/D7042

phabricator: add the DiffChangeType and DiffFileType constants These are used in Phabricator change objects. There are more values but so far as I can tell we don't need them. Differential Revision: https://phab.mercurial-scm.org/D7041

help: adding a help category to narrow and remotefilelog commands Differential Revision: https://phab.mercurial-scm.org/D6998

rust-cpython: drop self.borrow_mut() in favor of PySharedRef wrapper

rust-cpython: drop self.leak_immutable() in favor of PySharedRef wrapper

rust-cpython: add safe wrapper representing shared data borrowed from PyObject PySharedRef is a tempoary wrapper around PySharedRefCell. It provides safe functions for each shared data. $shared_accessor implements a safe method to construct PySharedRefCell. This allows us to add more than once PySharedRefCell to a Python object.

rust-cpython: move $leaked struct out of macro It wasn't easy to hack the $leaked struct since errors in macro would generate lots of compile errors. Let's make it a plain struct so we can easily extend it. PyLeakedRef keeps a more generic PyObject instead of the $name struct since it no longer has to call any specific methods implemented by the $name class. $leaked parameter in py_shared_iterator!() is kept for future change.

rust-cpython: store leaked reference to PySharedState in $leaked struct I want to move it out of the macro, and allow multiple sharable objects per PyObject.

rust-cpython: mark PySharedState as Sync so &'PySharedState can be Send The goal is to store &'static PySharedState in $leaked struct, which allows us to move the $leaked struct out of the macro. Currently, it depends on inner.$data_member(py), which can't be generalized. PySharedState is Sync because any mutation or read operation is synchronized by the Python GIL, py: Python<'a>, which should guarantee that &'PySharedState can be sent to another thread.

rust-cpython: move borrow_mut() to PySharedRefCell PySharedRefCell() will host almost all py_shared public functions. This change is the first step. borrow_mut() can be safely implemented since PySharedRefCell knows its inner object is managed by its own py_shared_state.

rust-cpython: move py_shared_state to PySharedRefCell object The goal of this series is to encapsulate more "py_shared" thingy and reduce the size of the macro, which is hard to debug. Since py_shared_state manages the borrowing state of the object owned by PySharedRefCell, this change makes more sense. If a PyObject has more than one data to be leaked into Python world, each PySharedState should incref the parent PyObject, and keep track of the corresponding borrowing state.

py3: decode username and password before SMTP login smtplib.SMTP.login() requires str on Python 3. For 'password', we only need to decode when value comes from config as getpass() returns the correct type already.

py3: use socket.makefile() instead of dropped smtplib.SSLFakeFile The latter (undocumented internal) class got removed in Python 3.3. Use socket.makefile() as suggested: https://docs.python.org/3/whatsnew/3.3.html#porting-python-code

py3: call SMTP.docmd() with an str Otherwise, this always returns (502, '5.5.2 Error: command not recognized').

py3: call SMTP.has_extn() with an str Passing a bytes on Python 3 always returns False, thus starttls is not properly detected.

py3: fix sorting of obsolete markers when building bundle Last item of marker tuple (parents) is either None or tuple. Comparison thus fails on Python 3 with: TypeError: '<' not supported between instances of 'tuple' and 'NoneType' Fixing this by coercing None to the empty tuple when sorting markers in exchange._getbundleobsmarkerpart().

sidedata: rename the configuration option to `exp-use-side-data` We don't want setup configured to use the final version that would end using the experimental one while using and older version. Differential Revision: https://phab.mercurial-scm.org/D7040

py3: drop ui.flush() during interactive patch filtering With previous changeset fixing line buffering on stdout, this is no longer needed.

py3: keep stdout as defined by pycompat in procutil According to https://docs.python.org/3/library/functions.html#open, it's not possible to use 1 as buffering argument value in binary mode. This is probably why line buffering does not work well in python3. On the other hand, sys.stdout.buffer appears to be line-buffered already on python3. So by not replacing it, there should be no behavior change. This fixes buffering issue in "hg email" (confirmation prompt shown before information to be confirmed).

notify: cast hash to bytes This is needed to avoid a str/bytes mismatch when interpolating a line or 2 later. Differential Revision: https://phab.mercurial-scm.org/D7021

push: support config option to require revs be specified when running push Differential Revision: https://phab.mercurial-scm.org/D6989

py3-discovery: using plain str in stats dict rust-cpython converts automatically from Rust strings to the appropriate `str` for the target Python version. Insisting on discovery stats dict keys to be bytes hence breaks the process (this is spotted by test-setdiscovery.t). Now that byteify-strings has been run on the entire codebase, and the import transformer is not there any more, the simplest fix is to make the keys plain str again. Another possible fix would be to forcefully convert to bytes in rust-cpython code, but that feels less natural, and would probably have to be reverted down the road. Differential Revision: https://phab.mercurial-scm.org/D7039

perf: fix `perfhelper-mergecopies` report of #changesets Same as `perfhelper-pathcopies`. The previous computation `<base>::<target>` was wrong, what we actually need is `::<target> - ::<base>`. This is now fixed.

perf: fix `perfhelper-pathcopies` report of #changesets The previous computation `<base>::<target>` was wrong, what we actually need is `::<target> - ::<base>`. This is now fixed

py3: use integer division in curseschunkselector.printstring() This fixes a crash when scrolling in curses UI when refresh() is called when a float value (namely 'self.firstlineofpadtoprint', taking its value indirectly from 'self.linesprintedtopadsofar').

crecord: drop duplicated set of firstlineofpadtoprint attribute The attribute is already set a couple of lines above, in curseschunkselector.__init__().

patchbomb: use mail.Generator alias for py2/py3 compat

py3: use email.generator.BytesGenerator in patch.split() This fixes test-import.t on python3. We add Generator alias in mail module to handle python2/python3 compatibility.

py3: only flush before prompting during interactive patch filtering Follows up on c9093ae8d6c4. It's enough to flush just before each prompt.

py3: add a missing b'' prefix in test extension for chg Differential Revision: https://phab.mercurial-scm.org/D7038

fuzz: new target to fuzz jsonescapeu8fast This code just feels complicated enough we should go ahead and give it a dedicated fuzzer: we've found bugs in similar things before. Differential Revision: https://phab.mercurial-scm.org/D7034

fuzz: new fuzzer for fncache-related functions Not all of these are strictly fncache-related, but they all have th same signature and similar-enough behavior that we may as well fuzz them together. No obvious bugs for once, but these felt like they were just complicated enough to cover. Differential Revision: https://phab.mercurial-scm.org/D7033