hg-core: remove unneeded trait now that we support Rust 1.52+

rust: remove newly redundant `use` statements with the 2021 edition prelude https://doc.rust-lang.org/edition-guide/rust-2021/prelude.html

rust: move all crates in the main workspace to edition 2021 We've changed our minimum Rust version to 1.61.0 in the previous patch, and edition 2021 predates that version.

rust: upgrade supported Rust toolchain version A few months ago¹, a decision was made to move the Rust toolchain target to whatever Debian Testing was tracking. I didn't have the bandwidth to act on it until now. This is starting to be even more problematic than before, now that edition 2021 is out. The CI has been updated to track the current Debian testing version, 1.61.0. [1] https://lists.mercurial-scm.org/pipermail/mercurial-packaging/2022-April/000338.html

branching: merge stable into default

tests: fix test-sparse-revlog This one is not covered by the CIbecause I requires an expensive artifact to be cached. So it goes out of think on regular basis (we should fix that…) The test ouput was affected by e706bb41fdb3 as we filtering now happens sooner, removing for the output.

delta-find: adjust the moment when we mark something as "tested" In a coming change, not all elements of `group` might get tested. So we need to have more control about when a revision is actually added to the `tested` set. So we move to a more verbose (and more fragile) version.

delta-find: rename a variable for clarity the index in the delta-chain is also the snapshot depth. So we rename the variable for clarity.

delta-find: small documentation update This is not a 1-1 mapping, but a 1-n mapping. Lets make the associated comment clearer.

delta-find: move pre-filtering with other pre-filtering logic This is more consistent and will help use to be in a clean state before dealing with the "too large group" issue. As a side effect, the debug output now skip some useless cases, making it more useful.

delta-find: expand a function definition and call before extendin it This make the next changeset more compact.

debug: add an option to display statistic about a unbundling operation This will helps a lot to understand how the bundling decision have actually impacted pull/unbundle on the other side.

debug: add an option to display statistic about a bundling operation This will helps a lot to understand how the bundling decision might impact pull/unbundle on the other side.

delta-find: add debug information about reuse of cached data This will help us to understand the behavior of find-delta during a pull.

cffi: fix a bytes vs str issue on macOS when listing directories This code hasn't been touched in recent years, and the other implementation return bytes for the filename, so I assume this is a holdover from the py2 days. I was unable to test it on mac though, because the `_osutil` import failed.

typing: fix the typehint for `skip` arg on `osutil.listdir()` to be optional The cffi and pure implementations both default to `None`, and the args are parsed in C with `"y#|OO:listdir"`, with the C variable initialized to NULL.

typing: add type hints to mpatch implementations Again, using `merge-pyi` to apply the stubs in cext and then manually type the private methods. The generated stub without these hints inferred very little, and the stuff it did was wrong.

typing: add type hints to bdiff implementations Not super important code, but this was an exercise in using `merge-pyi` to fold type stubs back into the code on something small. The cext stubs don't seem to be getting used (at least the only thing in `.pytype/pyi/mercurial/cext` after a run generating the stubs is `__init__.pyi`), so maybe this will help some.

cffi: adjust the list returned by bdiff.blocks to never have a None entry This was flagged by pytype after merging the corresponding bdiff.pyi in cext: File ".../mercurial/cffi/bdiff.py", line 44, in blocks: bad return type [bad-return-type] Expected: List[Tuple[int, int, int, int]] Actually returned: List[None] AFAICT, all callers immediately unpack the tuple into 4 variables, so a `None` entry would simply crash if they aren't all overwritten. As long a `count` and the link list are consistent, this shouldn't be a problem. This placates both pytype and PyCharm (which complained about the `i` in `rl[i]` having the wrong type with the old code).

typing: fix a syntax error in mercurial/cext/bdiff.pyi I noticed because `merge-pyi` on the non-cext implementations with this file as input skipped the return type for this. `pytype-single --parse-pyi` confirmed it was a problem.

debug-delta-find: add a --source option This will help us to understand the delta-find operation in different situations.

demandimport: ensure lazyloaderex sets loader attributes (issue6725) Adds test capturing missed expectation.

shelve: add test for Shelf.changed_files

shelve: add Shelf.changed_files for resolving changed files in a plugin

branching: merge stable into default

configitems: enable pullbundles by default The use of pullbundles is already protected: they are only used when a pullbundles.manifest file is created on the server. Having an additional flag doesn't really make sense and can confuse users (as indicated in the mercurial mailing list on the topic "Can't get pull-bundles working").

typing: add basic type hints to stringutil.py

vfs: make the default opener mode binary The default was already binary for `abstractvfs`, and the `vfs` implementation adds binary mode if the caller didn't supply it. Therefore, it should be safe for all vfs objects (and I don't think we want text reads anyway).

typing: add basic type hints to vfs.py Again, there's a lot more that could be done, but this sticks to the obviously correct stuff that is related to primitives or `vfs` objects. Hopefully this helps smoke out more path related bytes vs str issues in TortoiseHg. PyCharm seems smart enough to apply hints from annotated superclass functions, but pytype isn't (according to the *.pyi file generated), so those are annotated too. There was some discussion about changing the default path arg from `None` to `b''` in order to avoid the more verbose `Optional` declarations. This would be more in line with `os.path.join()` (which rejects `None`, but ignores empty strings), and still not change the behavior for callers still passing `None` (because the check is `if path` instead of an explicit check for `None`). But I didn't want to hold this up while discussing that, so this documents what _is_.

util: implement `writelines()` on atomictempfile With typehints on the vfs objects, pytype will flag this: FAILED: /mnt/c/Users/Matt/hg/.pytype/pyi/mercurial/patch.pyi /usr/bin/python3.8 -m pytype.single --imports_info /mnt/c/Users/Matt/hg/.pytype/imports/mercurial.patch.imports --module-name mercurial.patch -V 3.7 -o /mnt/c/Users/Matt/hg/.pytype/pyi/mercurial/patch.pyi --analyze-annotated --nofail --quick /mnt/c/Users/Matt/hg/mercurial/patch.py File "/mnt/c/Users/Matt/hg/mercurial/patch.py", line 535, in writerej: No attribute 'writelines' on mercurial.util.atomictempfile [attribute-error] In Union[ mercurial.util.atomictempfile, mercurial.vfs.checkambigatclosing, mercurial.vfs.delayclosedfile, mercurial.windows.fdproxy, mercurial.windows.mixedfilemodewrapper ] It's not a real problem there (atomictempfile is only created by passing different args), but it's reasonable for this to implement the function and behave like a normal file. There are other functions missing that can be added if/when needed.

typing: add basic type hints to localrepo.py There's a lot more that could be done, but this sticks to the obviously correct stuff that is either related to existing imports or primitives. Hopefully this helps smoke out more path related bytes vs str issues in TortoiseHg. I'm avoiding the interfaces for now, because they seem to confuse pytype and/or PyCharm. It might be worth typing the return of `makelocalrepository` to `localrepository`, but that leaks an implementation detail, so that can be revisited later.

check-code: drop the check for whitespace around named parameters This check flags py3 annotations of named parameters, because `black` adds spaces around the assignment in this case. Since the chosen formatter has opinions (and pylint also wants the space in the case of annotations), drop the check so we can use py3 annotations.

rust-status: query fs traversal metadata lazily Currently, any time the status algorithm needs to read a directory from the filesystem (because the stat-only optimization is not available), it also stats each directory entry eagerly. Stat'ing the entries is only needed in a few cases (like when checking the mtime of a directory for caching): this patch creates a wrapper struct `DirEntry` that only stats the directory entry it represents when needed. Excerpt of an `strace` before this change on Mozilla Central: ``` openat(AT_FDCWD, ".", O_RDONLY|O_NONBLOCK|O_CLOEXEC|O_DIRECTORY) = 3 newfstatat(3, "", {st_mode=S_IFDIR|0755, st_size=3540, ...}, AT_EMPTY_PATH) = 0 getdents64(3, 0x55dc970bd440 /* 139 entries */, 32768) = 5072 statx(3, ".hg", AT_STATX_SYNC_AS_STAT|AT_SYMLINK_NOFOLLOW, STATX_ALL, {stx_mask=STATX_ALL|STATX_MNT_ID, stx_attributes=0, stx_mode=S_IFDIR|0755, stx_size=772, ...}) = 0 [... 135 other successful `statx` calls] getdents64(3, 0x55dc970bd440 /* 0 entries */, 32768) = 0 close(3) = 0 ``` After this change: ``` openat(AT_FDCWD, ".", O_RDONLY|O_NONBLOCK|O_CLOEXEC|O_DIRECTORY) = 3 newfstatat(3, "", {st_mode=S_IFDIR|0755, st_size=3540, ...}, AT_EMPTY_PATH) = 0 getdents64(3, 0x561567c10190 /* 139 entries */, 32768) = 5072 getdents64(3, 0x561567c10190 /* 0 entries */, 32768) = 0 close(3) = 0 ```

rust-status: make `DirEntry` attributes clearer

tests: remove non-python3 line matching and tests block We don't support Python2 anymore

pywatchman: remove obsolete comments about importing from future See 6000f5b25c9b.

demandimport: remove an obsolete comment about importing from future See 6000f5b25c9b.

mr-template: wrap the instructions inside a comment block At least in preview mode, this hides the text so the user doesn't have to delete it. It's still visible in edit mode, so the user sees it.

revlog: use the user facing filename as the display_id for filelogs I had trouble isolating some LFS blob corruption detected by `hg verify` because the traceback referenced a file, but with the `data/` prefix in the `.hg/store` path, so it couldn't be located with the `file()` revset: ``` Traceback (most recent call last): File "/mnt/d/mercurial/mercurial/revlog.py", line 3209, in verifyintegrity _verify_revision(self, skipflags, state, node) File "/mnt/d/mercurial/hgext/lfs/wrapper.py", line 246, in _verify_revision orig(rl, skipflags, state, node) File "/mnt/d/mercurial/mercurial/revlog.py", line 158, in _verify_revision rl.revision(node) File "/mnt/d/mercurial/mercurial/revlog.py", line 1816, in revision return self._revisiondata(nodeorrev, _df) File "/mnt/d/mercurial/mercurial/revlog.py", line 1870, in _revisiondata self.checkhash(text, node, rev=rev) File "/mnt/d/mercurial/mercurial/revlog.py", line 1996, in checkhash % (self.display_id, pycompat.bytestr(revornode)) mercurial.error.RevlogError: integrity check failed on data/EXE/PPC/shrinksrec.exe:0 ``` (I'm a little surprised it resulted in a stacktrace instead of just a message, but that's a different issue. I'm also not sure how to trigger the simplestore case, since IIUC, it's also a revlog based store.) It's not clear how to handle the changelog and manifest (because the user doesn't interact with them as a file), so those cases are left alone. The other thing that would be nice to improve somehow is to indicate that the ":0" is a revlog revision, not the changeset revision that users are used to. I'm not sure how to handle the "or node" part though.

revlog: drop an unused variable assignment It's assigned again 2 lines later.

lfs: improve an exception message for blob corruption detected on transfer The message about the server crash originated in 0ee0a3f6a990 (after support for serving blobs was added), but was copied from the Facebook repo that forked prior to server side support. Therefore, this message only displayed in their client, so it was safe to assume the server crashed. But that was never the case for vanilla Mercurial, as I saw this in a server log. Also, display the blob reference so that it's easier to figure out where the problem was when a bunch of blobs are transferred at once.

branching: merge stable into default

Added signature for changeset 8830004967ad

Added tag 6.3.3 for changeset 8830004967ad

relnotes: add 6.3.3

typing: add typehints to mercurial/diffutil.py Lack of typehints here caused the fact that TortoiseHg was passing str instead of bytes as the key in `opts` to be missed, resulting in shelf corruption in cases where `diff.git` is required.

patchbomb: respect the `--git` option I *think* this is the only diffopt exposed on the command line. TortoiseHg had a similar issue creating diffopts, and this was caught by type hints in the next commit.

rhg: remember the inode of .hg/dirstate This allows us to detect changes of `.hg/dirstate`, which is either the full dirstate (in dirstate-v1) or the docket file (v2) without relying on data inside the file. It only works on UNIX systems. This fixes a race condition for dirstate-v1 (as demonstrated by the test changes) and adds a confortable layer of sanity for dirstate-v2.

rust-dirstate-v2: don't write dirstate if data file has changed This fixes the following race: - process A reads the dirstate - process B reads and writes the dirstate - process A writes the dirstate This either resulted in losing what process B had just written or a crash because the `uuid` had changed and we were trying to write to a file that doesn't exist. More explanations inside. This doesn't fix the issue for dirstate-v1, a later patch addresses it.

rust-dirstate: remember the data file uuid dirstate was loaded with This will be used in the next patch to fix a race condition.

dirstate: set identity whenever we read the dirstate's v2 docket The docket can be loaded outside of a full read (for exemple when pre-fetching parents), so the current code would read/set the identity after loading the data, opening a race condition: A0: first process docket is read B0: other process appends new data to the dirstate (and changes the docket) A1: first process sets the identity (based on pre-B content, but with post-B identity) A1: first process loads the dirstatemap from the data file A1: first process does not detect the race and overwrites the update from B.

dirstate: factor the identity setting code in the dirstate map We need it in more locations, so let us start factoring thing out first to make sure the same code is called everywhere. This bears some similarity with 85746485a4dd on default, but at a smaller scope and for a different purpose.

dirstate: simplify the dirstate's read race testing Now that most code behaves properly, we can simplify the expected matching.

dirstate: deal with read-race for pure rust code path (rhg) If we cannot read the dirstate data, this is probably because a writing process wrote it under our feet. So refresh the docket and try again a handful of time.

dirstate: deal with read-race for python code using rust object If we cannot read the dirstate data, this is probably because a writing process wrote it under our feet. So refresh the docket and try again a handful of time.

dirstate: deal with read-race for pure python code If we cannot read the dirstate data, this is probably because a writing process wrote it under our feet. So refresh the docket and try again a handful of time.

dirstate: abstract the reading of the data file in v2 in a method We will need more changes to avoid some race conditions during read, so we first isolate the simple logic before making it more complicated.

dirstate: add append/new-file variants in the dirstate's read race tests This covers more ground and finds more bugs. At that point I gave up on making things as `known-bad-output` / `missing-correct-output` as this gets too messy.

dirstate: add a synchronisation point in the middle of the read This will be useful to test some more race conditions around dirstate.

dirstate: add v1-v2 variants to the dirstate's read race tests More cases mean different issues.

dirstate: check dirstate race condition around status More problems to solve.

dirstate: check dirstate race condition around update More problems to solve.

dirstate: check dirstate race condition around commit Once in a while, rhg is the only one to behave right here.

dirstate: initial creation of a test file to check dirstate race read More problems to solve… yeah ! (I guess)

dirstate: add a synchronisation point before doing a full dirstate read This will be useful to test some race conditions around the dirstate.

rust-repo: move dirstate-v2 opening to a separate method The next changeset will make changes to this logic, it helps to have it in order first.

rhg: fix race when an ambiguous file is deleted on disk There are two places in the status code where we handle files whose status we are unsure of based off of metadata alone: this one is the first one to actually disambiguate, and the second one is later in the code (but updated in the previous commit) for files that are actually clean to update the dirstate. Since there is a chance that the contents have changed between those two moments, we need to stat the files again, since re-using the old stat could lie about the clean state of the file.

rhg: fix race when a fixup file is deleted on disk See next changeset for the other race in the same kind of logic and why there are two different places.

dirstate: test a `hg status` raced by a `hg remove` This shows that `rhg` is misbehaving here.

dirstate: tests racing status with both dirstate-v2 append and rewrite The way the racing process touches the dirstate results in different challenges for the raced process. We now test each variant in the `test-dirstate-status-race.t` tests.

dirstate-v2: add devel config option to control write behavior This will help us to write predictable tests checking behavior in each case.

dirstate: use more than a bool to control append behavior When writing dirstate-v2, we might either append to the existing file, or create a new file. We are about to introduce some configuration to control this behavior. As a prelude, we change the current way the behavior was automatically controlled to make the change smaller/clearer.

dirstate: cover each dirstate version when testing for status race Previously we were only testing it with the default (dirstate-v1 currently). Now we explicitly test each variant.

dirstate: test a `hg status` raced by another `hg status` This shows that `rhg` is misbehaving here.

dirstate: test a `hg status` raced by a `hg update` This shows that `rhg` is misbehaving here.

dirstate: test a `hg status` raced by a `hg commit` This shows that `rhg` is misbehaving here.

dirstate: test a `hg status` raced by a `hg add` This shows that `rhg` is misbehaving here.

dirstate: add a way to test races happening during status We add the `devel.sync.status.pre-dirstate-write-file` config option to easily test what happens when other operations happen during the window where `hg status` is done working but has not updated the cache on disk yet. We introduce the framework for testing such races too, actual tests will be added in the next changesets. For now the test is only checking dirstate-v1. We will extend the test coverage later too. Check test documentation for details. Code change from Raphaël Gomès <rgomes@octobus.net> Test change from Pierre-Yves David <pierre-yves.david@octobus.net>

testing: introduce util function to synchronize concurrent commands on files This is an extension of mechanisms that the tests have been using for a while. To be able to also control the execution in Rust, we introduce utility to perform such `wait_on_file` logic based on some configuration value. This will be used in the tests introduced in the next changesets.

dirstate: add some debug output when writing the dirstate This helps debugging Mercurial behavior in the dirstate-v2 case.

run-tests: make it possible to nest conditionals This is not that hard to implement and makes our life easier on a regular basis.

dirstate-v2: don't mmap the data file when on NFS `mmap` on NFS will trigger a SIGBUS when the mmap'ed file is deleted, which wouldn't work in our case. Also, the performance advantage of using mmap on NFS is debatable at best.

rust-dirstate: trace append/no append to help debugging

rust: add debug log about skipping dirstate update

test-dirstate: use more robust method to trigger a data-file append The previous method was fragile and somewhat flaky on fast machines.

transaction: tests we don't overwrite bookmark activation on abort We actually do not! Great.

transaction: tests we don't overwrite updates on abort spoiler: we do… /o\

transaction: tests we don't overwrite branch changes on abort We actually do not! Great. …Why are doing a backup of the `branch` files at transaction creation then‽

transaction: tests we don't overwrite tracking to changed file on abort spoiler: we do…

transaction: the base of a new test file checking transaction abort issue See inline documentation for details. See other changesets for actual cases.

setup: support building from an ongoing merge Before this change the two parents from the merge would duplicate some command output and modify some others in a way python 3.11 chokes on.

rust: upgrade minimum `rayon` dependency The 1.6.0 and 1.6.1 releases fixes a soundness issue and a performance issue that are both important to us, since both correctness and performance are paramount in the Rust extensions.

test-chg: use a different log to avoid flakyness The test was deleting the log file to start anew. However a trailing working process might still be alive at this time, and recreate the very same log on exit. We see the trace of such worker in the expected content of server.log (see the trace modified by this patch). This is flaky because we don't know *when* the worker will write to the file and there is a race with the `hg init cached` command. A much simpler and reliable way to start anew without having such race is… to write to a different log file. No reuse → no conflict, no conflict → no race, no race → no flakiness.

dirstate-v2: complain early on docket name collision The alternative is that the dirstate gets deleted so the corruption persists and is hard to investigate. This happened to me in tests, where the dirstate names are taken from file, since the file got reverted. I expect this can also happen in prod with non-trivial probability (1/4 billion).

setup: further improve the error path for version retrieval This is a new take at the problem that 8d390a13474d tried to tackle. There was two issues after that previous improvement: - the 0.0+ version could survive a bit too long and reaching the installer version and staying there. - multiple use case where still failing. So the new code is better at: - always succeeding when running `make local` so that we can bootstrap a local version - no using that fallback outside of `make local` to avoid distribution of version with the buggy version number. The setup.py is a gigantic pile of spaghetti code, to the point where pastafarian pilgrim started knocking at its core. However I refrained from cleaning that up since the more to a `setup.cfg` means this code should be deleted soon™.

dirstate: handle missing backup file on restoration This is the stable counter part to e358f6e0e50e. Since 6.4 will stop writing undo.dirstate in some case (actually… at all), a transaction created with 6.4 and recover/rolledback with 6.3 need to work to a certain degreee. This changeset add the necessary bits so that we don't get a traceback from 6..3 in this cases.