blackbox: fix type error on log rotation on read-only filesystem Grepping around, the code uses either encoding.strtolocal or stringutil.forcebytestr in this situation. No idea which is best. Differential Revision: https://phab.mercurial-scm.org/D10293

rust: Remove use of `py.eval()` The previous Rust code allocated an intermediate `Vec`, converted that to a Python list, then used `eval` to run Python code that converts that list to a Python set. rust-cpython exposes Rust bindings for Python sets, let’s use that instead to construct a set directly. Differential Revision: https://phab.mercurial-scm.org/D10328

rust: Remove the compile-time 'dirstate-tree' feature flag This code has compiler errors since it is not built on CI and nobody has been working on it for some time. We (Octobus) are still pursuing status optimizations based on a tree data structure for the dirstate, but upcoming patches will use a run-time opt-in instead of compile-time, so that at least corresponding Rust code keeps compiling when other changes are made. Differential Revision: https://phab.mercurial-scm.org/D10329

procutil: avoid using os.fork() to implement runbgcommand We ran into the following deadlock: - some command creates an ssh peer, then raises without explicitly closing the peer (hg id + extension in our case) - dispatch catches the exception, calls ui.log('commandfinish', ..) (the sshpeer is still not closed), which calls logtoprocess, which calls procutil.runbgcommand. - in the child of runbgcommand's fork(), between the fork and the exec, the opening of file descriptors triggers a gc which runs the destructor for sshpeer, which waits on ssh's stderr being closed, which never happens since ssh's stderr is held open by the parent of the fork where said destructor hasn't run Remotefilelog appears to have a hack around this deadlock as well. I don't know if there's more subtlety to it, because even though the problem is determistic, it is very fragile, so I didn't manage to reduce it. I can imagine three ways of tackling this problem: 1. don't run any python between fork and exec in runbgcommand 2. make the finalizer harmless after the fork 3. close the peer without relying on gc behavior This commit goes with 1, as forking without exec'ing is tricky in general in a language with gc finalizers. And maybe it's better in the presence of rust threads. A future commit will try 2 or 3. Performance wise: at low memory usage, it's an improvement. At higher memory usage, it's about 2x faster than before when ensurestart=True, but 2x slower when ensurestart=False. Not sure if that matters. The reason for that last bit is that the subprocess.Popen always waits for the execve to finish, and at high memory usage, execve is slow because it deallocates the large page table. Numbers and script: before after mem=1.0GB, ensurestart=True 52.1ms 26.0ms mem=1.0GB, ensurestart=False 14.7ms 26.0ms mem=0.5GB, ensurestart=True 23.2ms 11.2ms mem=0.5GB, ensurestart=False 6.2ms 11.3ms mem=0.2GB, ensurestart=True 15.7ms 7.4ms mem=0.2GB, ensurestart=False 4.3ms 8.1ms mem=0.0GB, ensurestart=True 2.3ms 0.7ms mem=0.0GB, ensurestart=False 0.8ms 0.8ms import time for memsize in [1_000_000_000, 500_000_000, 250_000_000, 0]: mem = 'a' * memsize for ensurestart in [True, False]: now = time.time() n = 100 for i in range(n): procutil.runbgcommand([b'true'], {}, ensurestart=ensurestart) after = time.time() ms = (after - now) / float(n) * 1000 print(f'mem={memsize / 1e9:.1f}GB, ensurestart={ensurestart} -> {ms:.1f}ms') Differential Revision: https://phab.mercurial-scm.org/D9019

share: store relative share paths with '/' separators I created a relative share in Windows and tried to use it in WSL, and it failed: abort: .hg/sharedpath points to nonexistent directory /mnt/c/Users/Matt/hg-review/.hg/..\..\hg\.hg Use `normpath` on the read side so that the code has the usual Windows style paths it always had (I don't think that matters much), but it also eliminates the directory escaping path components in the case where the path is printed. This will not fix repositories that have already been created, but it's trivial enough to hand edit the file to correct it. Differential Revision: https://phab.mercurial-scm.org/D10330

unit-tests: Fix `cargo test` on 32-bit platforms Fixes https://bz.mercurial-scm.org/show_bug.cgi?id=6506 This makes `IndexEntryBuilder::build`, which is only used in unit tests, use `u32` or `u64` instead of platform-dependent `usize` when packing binary data to be used at test input. To run Rust unit tests in 32-bit mode in a x86-64 environment, use: rustup target add i686-unknown-linux-gnu # Once (cd rust && cargo test --target i686-unknown-linux-gnu) Differential Revision: https://phab.mercurial-scm.org/D10351

rename: add --forget option and stop suggesting `hg revert` for undoing Differential Revision: https://phab.mercurial-scm.org/D10355

win32: enable legacy I/O mode to fix missing pager output on Windows with py3 The equivalent interpreter option is set by wrapper.exe, but this *.bat file is what gets installed in a venv. Without this mode, any command that spins up a pager has no output, unless the pager is explicitly disabled. The variable is set inside the `setlocal` scope to keep it from leaking into the environment after the bat file exits. We should probably still figure out how to ship a compiled hg.exe when installing with `pip`, because the binary does other things like enable long filename support. But this avoids the dangerous and confusing lack of output in the meantime. Differential Revision: https://phab.mercurial-scm.org/D10354

persistent-nodemap: enable the feature by default when using Rust As discussed at the 5.6 sprint, we can make it enabled by default, but only for Rust installation. Differential Revision: https://phab.mercurial-scm.org/D9765

persistent-nodemap: disable it for test-wireproto-content-redirects.t This is not relevant for the test and this will avoid a lot of variations. Differential Revision: https://phab.mercurial-scm.org/D10338

persistent-nodemap: disable it for test-wireproto-command-capabilities.t This is not relevant for the test and this will avoid a lot of variations. Differential Revision: https://phab.mercurial-scm.org/D10337

persistent-nodemap: disable it unconditionally for test-wireproto-caching.t This is not relevant for the test and this will avoid a lot of variations. Differential Revision: https://phab.mercurial-scm.org/D10336

persistent-nodemap: disable it unconditionally for test-ssh-proto-unbundle.t This is not relevant for the test and this will avoid a lot of variations. Differential Revision: https://phab.mercurial-scm.org/D10335

persistent-nodemap: disable it unconditionally for test-ssh-proto.t This is not relevant for the test and this will avoid a lot of variations. Differential Revision: https://phab.mercurial-scm.org/D10334

persistent-nodemap: disable it unconditionally for test-share-safe.t This is not relevant for the test and this will avoid a lot of variations. Differential Revision: https://phab.mercurial-scm.org/D10333

persistent-nodemap: disable it unconditionally for test-http-protocol.t This is not relevant for the test and this will avoid a lot of variations. Differential Revision: https://phab.mercurial-scm.org/D10332

persistent-nodemap: disable it unconditionally for test-http-bad-server.t This is not relevant for the test and this will avoid a lot of variations Differential Revision: https://phab.mercurial-scm.org/D10331

clang-format: run the formatter on mercurial/cext/revlog.c This fix `test-check-clang-format.t` that has been complaining for a while. Differential Revision: https://phab.mercurial-scm.org/D10327

revlog-compression: use zstd by default (if available) As see in changeset bb271ec2fbfb, zstd is 20% to 50% faster for reading and writing. Use take advantage of the new config behavior to try zstd by default, falling back to zlib is zstd is not available on that plateform. Differential Revision: https://phab.mercurial-scm.org/D10326

rhg: make rhg recognise it supports zstd compression for revlogs It already did, but was not aware of it. Differential Revision: https://phab.mercurial-scm.org/D10324

match: convert O(n) to O(log n) in exactmatcher.visitchildrenset When using narrow, during rebase this is called (at least) once per directory in the set of files in the commit being rebased. Every time it's called, we did the set arithmetic (now extracted and cached), which was probably pretty cheap but not necessary to repeat each time, looped over every item in the matcher and kept things that started with the directory we were querying. With very large narrowspecs, and a commit that touched a file in a large number of directories, this was slow. In a pathological repo, the rebase of a single commit (that touched over 17k files, I believe in approximately as many directories) with a narrowspec that had >32k entries took 8,246s of profiled time, with 5,007s of that spent in visitchildrenset (transitively). With this change, the time spent in visitchildrenset is less than 34s (which is where my profile cut off). Most of the remaining time was network access due to our custom remotefilelog-based setup not properly prefetching. Differential Revision: https://phab.mercurial-scm.org/D10294

exthelper: improve docs to indicate what module vars are needed I recently tried creating an extension "from scratch" using exthelper, and it wasn't obvious that you needed these. I believe that a careful reading of one of the comments would tell you that they were required, but it's easy to miss and having the examples be "complete" is helpful. Differential Revision: https://phab.mercurial-scm.org/D10295

tests: restore the ability to run `black` on Windows The hghave test for black silently stopped working with the change in 08fd76a553c9. This was the output of what it was hitting when run in the test environment: Traceback (most recent call last):\r (esc) File "c:\\users\\matt\\appdata\\local\\programs\\python\\python39\\lib\\runpy.py", line 197, in _run_module_as_main\r (esc) return _run_code(code, main_globals, None,\r (esc) File "c:\\users\\matt\\appdata\\local\\programs\\python\\python39\\lib\\runpy.py", line 87, in _run_code\r (esc) exec(code, run_globals)\r (esc) File "c:\\Users\\Matt\\AppData\\Roaming\\Python\\Python39\\Scripts\\black.exe\\__main__.py", line 4, in <module>\r (esc) File "C:\\Users\\Matt\\AppData\\Roaming\\Python\\Python39\\site-packages\\black\\__init__.py", line 70, in <module>\r (esc) CACHE_DIR = Path(user_cache_dir("black", version=__version__))\r (esc) File "C:\\Users\\Matt\\AppData\\Roaming\\Python\\Python39\\site-packages\\appdirs.py", line 293, in user_cache_dir\r (esc) path = os.path.normpath(_get_win_folder("CSIDL_LOCAL_APPDATA"))\r (esc) File "C:\\Users\\Matt\\AppData\\Roaming\\Python\\Python39\\site-packages\\appdirs.py", line 481, in _get_win_folder_with_pywin32\r (esc) dir = shell.SHGetFolderPath(0, getattr(shellcon, csidl_name), 0, 0)\r (esc) pywintypes.com_error: (-2147024893, '$ENOTDIR$.', None, None)\r (esc) [1] Differential Revision: https://phab.mercurial-scm.org/D10310

setup: copy python3.dll next to hg.exe when building on Windows for hgext.git I thought I took care of this already, but it must have been that I just manually copied the file over locally when debugging why the pygit2 library wasn't loading. The problem with that is what was copied over was from py38, and then running a py39 build hard crashed when the extension was loaded. Differential Revision: https://phab.mercurial-scm.org/D10301

tests: stablize test-hook.t on Windows Apparently, hooks can't run `echo` directly, even from MSYS. Differential Revision: https://phab.mercurial-scm.org/D10300

tests: handle Windows file separator differences in test-config.t Differential Revision: https://phab.mercurial-scm.org/D10299

tests: update the detailed exit codes for icasefs gated tests The fact that there's already a detailed exit code after the last change here in test-casecollision-merge.t gives me some pause, but maybe it was found and changed manually? Differential Revision: https://phab.mercurial-scm.org/D10298

tests: update the detailed exit codes in test-phabricator.t I'm guessing this has slipped through because most people don't have the necessary `pytest-vcr` package installed. Differential Revision: https://phab.mercurial-scm.org/D10297

tests: avoid using rebaseskipobsolete=0 I'm about to delete the `rebaseskipobsolete` config. This patch updates a test to get hidden commits into the rebase state by using `hg debugobsolete` instead of setting `rebaseskipobsolete=0`. Differential Revision: https://phab.mercurial-scm.org/D10270

rebase: when using --keep, don't care about pruned commits or divergence `hg rebase --keep` creates duplicate commits (not successors), so I was surprised that it still skips pruned commits and errors out if it "would cause divergence" (it wouldn't). I guess this was just an oversight. We didn't have any tests for it, so I also included that. Differential Revision: https://phab.mercurial-scm.org/D10269

rebase: remove duplicate initialization of a field `obsolete_with_successor_in_destination` is already initialized in the constructor. Differential Revision: https://phab.mercurial-scm.org/D10268

rust: bump rust-cpython version to 0.5.2 we need a newer version to define "property" on Rust defined object. Differential Revision: https://phab.mercurial-scm.org/D10308

revlog: directly use the Struct object for related operation The Struct object has all the piece we needs, so no need to duplicate information on the revlog itself. Differential Revision: https://phab.mercurial-scm.org/D10307

revlog: move the "index header" struct inside revlog.utils.constants The struct was previous called "version", but this is actually "version" + "flags". So header seems like a better name. The move to the `constants` module has the same motivation as the INDEX_ENTRY_V# ones. Differential Revision: https://phab.mercurial-scm.org/D10306

revlog: move the details of revlog "v2" index inside revlog.utils.constants the revlog module is quite large and this kind of format information would handy for other module. So let us start to gather this information about the format in a more appropriate place. We update various reference to this information to use the new "source of truth" in the process. Differential Revision: https://phab.mercurial-scm.org/D10305

revlog: move the details of revlog "v1" index inside revlog.utils.constants The revlog module is quite large and this kind of format information would handy for other module. So let us start to gather this information about the format in a more appropriate place. We update various reference to this information to use the new "source of truth" in the process. Differential Revision: https://phab.mercurial-scm.org/D10304

revlog: move the details of revlog "v0" index inside revlog.utils.constants the revlog module is quite large and this kind of format information would handy for other module. So let us start to gather this information about the format in a more appropriate place. Differential Revision: https://phab.mercurial-scm.org/D10303

revlog: add some comment in the header sections We are about to add more content so let us organise the existing content first. Differential Revision: https://phab.mercurial-scm.org/D10302

store: drop the `filefilter` argument to `_walk` No code use it anywhere. Dropping it will help replacing the function with something with a more precise semantic. Differential Revision: https://phab.mercurial-scm.org/D10314

store: document the `walk` method Differential Revision: https://phab.mercurial-scm.org/D10313

revlog: fix error about unknown compression format in py3 In py2, the error is something like: abort: unknown compression type 'x'! In py3, we get the following unhelpful message: abort: unknown compression type <memory at 0x7f4650b5cdc8>! Switch to something like: abort: unknown compression type 78! Differential Revision: https://phab.mercurial-scm.org/D10318

revlog-compression: fix computation of engine availability We don't just need the engine to be define, we need it to be available and able to do be used for revlog compression. Without this change, `zstd` could be selected as a viable option for repository creation on platform where it is not available. Differential Revision: https://phab.mercurial-scm.org/D10325

test: explicitly use zlib compression in tests/test-share-safe.t We need the implicit value to be explicit until we can change the default in some case. Differential Revision: https://phab.mercurial-scm.org/D10323

test: explicitly use zlib compression in tests/test-upgrade-repo.t We need the implicit value to be explicit until we can change the default in some case. Differential Revision: https://phab.mercurial-scm.org/D10322

test: explicitly use zlib compression in tests/test-repo-compengines.t We need the implicit value to be explicit until we can change the default in some case. Differential Revision: https://phab.mercurial-scm.org/D10321

remotefilelog: include file contents in bundles produced during strip `hg strip` and other things that use repair.strip (such as the narrow extension's `hg tracked --removeinclude`) will "save" some commits that have a higher revision number than the oldest commit we're stripping, but aren't actually descended from any of the commits that we're stripping. It saves them in a bundle, and then reapplies them to the repo. Remotefilelog doesn't generally participate in strip, it doesn't contribute files to either the backup bundle or the "saved" bundle, and doesn't adjust linknodes when commits are stripped. This can break things like push, which rely on the linknodes. This change makes it so that remotefilelog includes files in these bundles during strip operations. During reapplication, the files are reapplied from the bundle, and the linknode is properly updated. Differential Revision: https://phab.mercurial-scm.org/D10320

tests: add test-remotefilelog-strip.t to demonstrate an issue with linknodes ### Background Every time a commit is modified, remotefilelog updates the metadata for the file object to point to the new commit (I believe that this is different from non-remotefilelog hg, which leaves the linkrevs pointing to the obsolete commits; doing otherwise would involve changing data in the middle of revlogs). With `hg strip` (or other things that use repair.strip()), when you strip a commit that's not the tip of the revlog, there may be commits after it in revnum order that aren't descended from it and don't need to be (and shouldn't be) stripped. These are "saved" by strip in a bundle, and that bundle is reapplied after truncating the relevant revlogs. ### The problem Remotefilelog generally avoids being involved at all in strip. Currently, that includes even providing file contents to this backup bundle. This can cause the linknode to point to a changeset that is no longer in the repository. Example: ``` @ 3 df91f74b871e | | x 2 70494d7ec5ef |/ | x 1 1e423846dde0 |/ o 0 b292c1e3311f ``` Commits 1, 2, and 3 are related via obsolescence, and are description-only changes. The linknode for the file in these commits changed each time we updated the description, so it's currently df91f7. If I strip commits 1 and 3, however, the linknode *remains* df91f7, which no longer exists in the repository. Commit 70494d was "saved", stripped, and then reapplied, so it is in the repository (as revision 1 instead of 2 now), and was unobsoleted since the obsmarker was stripped as well. The linknode for the file should point to 70494d, the most recent commit that is in the repository that modified the file. Remotefilelog has some logic to handle broken linknodes, but it can be slow. We have actually disabled it internally because it's too slow for our purposes. Differential Revision: https://phab.mercurial-scm.org/D10319

mergestate: remove unused import Differential Revision: https://phab.mercurial-scm.org/D10291