vfs: raise NotImplementedError in abstractvfs._auditvfs This make sure that auditing is implemented.

vfs: handle _auditpath in proxyvfs Just forward the call to the underlying vfs.

vfs: fix proxyvfs inheritance The proxyvfs class is designed to overwrite some of the vfs logic. Yet, it did not use normal class inheritance. This is becoming an issue as `abstractvfs` method could take precedence over their `proxyvfs` version. We fix the inheritance chain to be as expected.

vfs: also audit rename Renaming through the vfs is not used in many places, and none of them seems to be a security risk. However, it is still worthwhile to run the auditing on rename file to perform developer-warning level checks.

vfs: add a `_auditpath` to abstract vfs We are about to make `rename` audit path. Since rename lives in the `abstractvfs` layer, we need it to be aware of auditing to some extent. The default implementation is no-op because multiple existing vfs are not using auditing at all right now (eg: fncachevfs).

vfs: makes all audited path relative Only auditing relative path helps the vfs warning logic.

debugupgraderepo: add a --no-backup mode The process has been around for a while and is pretty safe now. Having an automated way to clean up the old data is useful when running many different conversion to compare delta algorithm.

upgrade: add '-' in optimization name The older name `redeltaall` was hard to type and read. The newer form should be more user-friendly. We keep backward compatibility with the old form (at least for a while). Having to use different form depending on the version is very impractical and error prone.

revlog: add test case for _findsnapshots Testing the method directly is more robust.

revlog: use the native implementation of issnapshot In some sparserevlog case where a lot of the history has to be searched for a snapshot, the cost of issnashot cost becomes significant. The computation done by the method is fairly low level, a native implementation provide a very significant speedup. example affected manifest write before: 0.490375s after: 0.114989s (-76%)

revlog: add a native implementation of issnapshot This will be used in the next changesets

revlog: more efficient implementation for issnapshot We avoid multiple method calls and tuple creation, this provides a significant speedup in some case: example affected manifest write before: 0.815520s after: 0.487767s (-40%)

revlog: clarify the non sparse-revlog case in `issnapshot` If we are not using sparse-revlog, there is only one type of snapshot, the full one.

exthelper: switch to using the registrar merging method

registrar: add a method to merge registrar instances This provides sanity checking beyond simply merging the underlying dictionaries.

phabricator: ensure that the return of urlopener.open() is closed No problem observed, just an oversight noticed while reading documentation.

revlog: fix pure python slicing test when chain contains nullrev The revlog mock used in the test was not behaving the same as its C counterpart.

merge with stable

remotefilelog: fix bug in maybesparsematch returning alwaysmatcher The description of the method says that it should return None if sparse is not used in this repository; since sparse.matcher() returns alwaysmatcher if sparse is not enabled, I'm using that as the signal to return None here to preserve the previous behavior. Differential Revision: https://phab.mercurial-scm.org/D5487

procutil: correct spelling of uninterruptable -> uninterruptible Differential Revision: https://phab.mercurial-scm.org/D5488

help: show advanced, experimental and deprecated extensions with --verbose I noticed that phabricator and absorb weren't showing up, no matter what. There's a related problem where commands for these extensions *do* show up without --verbose. I'm not sure what the point of hiding the extensions, but not the commands is.

help: remove a duplicate category name entry for TOPIC_CATEGORY_NONE

phabricator: assign commands to help categories `phabsend` and `phabupdate` seem pretty close to `email`. `phabread` is a little less close, but can be trivially used to actually import patches.

largefiles: eliminate an unnecessary import of configitems Thanks to Yuya for pointing this out.

exthelper: add some examples for using registrar aliases Maybe it's my general lack of python knowledge, but how to use these would be way too obscure for me otherwise.

exthelper: reintroduce the ability to register filesets Same mechanism as revsets earlier in this series. The LFS extension is updated to provide test coverage.

exthelper: reintroduce the ability to register templates Same mechanism as revsets earlier in this series. The LFS extension is updated to provide test coverage. This also seems to make the test failure around issue6033 mentioned in 0a7f582f6f1f much less regular on Windows.

extensions: deprecate extsetup without a `ui` argument (API) 9.5 years should be enough time, but there were some tests for the old style still (which are now updated). Exthelper doesn't fallback to the old API, so this is for consistency. .. api:: The extension hook ``extsetup`` without a `ui` argument has been deprecated, and will be removed in the next version. Add a `ui` argument to avoid the deprecation warning.

largefiles: port revset registration to exthelper This tests the merge code that wasn't tested in the previous patch.

exthelper: reintroduce the ability to register revsets I think this is what Yuya and Boris agreed on.[1] This happens *after* the extsetup phase now (and after the _aftercallback notifications). But this is trivial, mergeable between exthelper instances, and doesn't need to have the extension name supplied when registering. The test needed updating so that extsetup() takes a `ui` argument, as exthelper isn't trying to be backward compatible with 1.3.1. [1] https://www.mercurial-scm.org/pipermail/mercurial-devel/2018-December/125888.html

largefiles: drop the uisetup module This is small enough to live in the __init__ module.

branches: add -r option to show branch name(s) of a given rev (issue5948) Differential Revision: https://phab.mercurial-scm.org/D5486

progress: avoid ui.configbool() lookup when progress bar is active Profiling revealed that the ui.configbool('progress', 'debug') during progress bar updates was consuming a significant amount of overhead. This commit adds an attribute on progress bar instances that caches this config option. The impact on `hg perfprogress` with default options is significant: before: ! wall 4.641942 comb 4.580000 user 4.210000 sys 0.370000 (best of 3) after: ! wall 1.948626 comb 1.950000 user 1.950000 sys 0.000000 (best of 5) After this change, profiling reveals that progress.progbar.progress() is now consuming ~73% of time. This change does not improve the execution time if the progress bar is disabled. We may want a more comprehensive solution for that case, as the progress bar won't be enabled in a number of scenarios (e.g. servers and processes not attached to an interactive TTY). I also think that overhead of ~2.0s for 1M updates is a bit high. I suspect further refactoring of the progress bar can significantly reduce overhead. I don't have plans to do this, however. Differential Revision: https://phab.mercurial-scm.org/D5408

largefiles: port wrapped functions to exthelper Things get interesting in the commit. I hadn't seen issue6033 on Windows, and yet it is now reproducible 100% of the time on Windows 10 with this commit. I didn't test Linux. (For comparison, after seeing this issue, I tested on the parent with --loop, and it failed 5 times out of over 1300 tests.) The strange thing is that largefiles has nothing to do with that test (it's not even mentioned there). It isn't autoloading run amuck- it occurs even if largefiles is explicitly disabled, and also if the entry in afterhgrcload() is commented out. It's also not the import of lfutil- I disabled that by copying the function into lfs and removing the import, and the problem still occurs. Experimenting further, it seems that the problem is isolated to 3 entries: exchange.pushoperation, hg.clone, and cmdutil.revert. If those decorators are all commented out, the test passes when run in a loop for awhile. (Obviously, some largefiles tests will fail.) But if any one is commented back in, the test fails immediately. I left one method related to wrapping the wire protocol, because it seemed more natural with the TODO. Also, exthelper doesn't support wrapping functions from another extension, only commands in another extension. I didn't try to figure out why rebase is both command wrapped and function wrapped.

largefiles: port commands to exthelper One subtle change here is that the purge, rebase and transplant extensions are wrapped in extsetup() instead of uisetup().

exthelper: support the option argument when registering a command Largefiles uses this 5th argument with push and pull, so this will be tested in the next commit. I assume the reason for unrolling and reforming the tuple in each finalxxxsetup() is to validate that something proper was passed in when registering. But it's better to explode when decorating than during the delayed actual registration.

largefiles: port configitems to exthelper It looks like dynamicdefault was referenced in a non-standard way.

remotefilelog: use repo.local() instead of isinstance()

exthelper: add a cautionary note about adding attributes to containers

exthelper: drop a compatibility shim with '^command' syntax This syntax was dropped in fa88170c10bb.

wireproto: in batch queries, support queries with immediate responses listkeys and pushkey return without querying the remote when the remote doesn't support such queries. Before this change, the batching code didn't handle this convention, resulting in this kind of error: $ hg pull ssh://user@dummy/repo1 -r tip -B a pulling from ssh://user@dummy/repo1 remote: ** Unknown exception encountered with possibly-broken third-party extension disable-lookup remote: ** which supports versions unknown of Mercurial. remote: ** Please disable disable-lookup and try your action again. remote: ** If that fixes the bug please report it to the extension author. remote: ** Python 2.7.15+ (default, Oct 2 2018, 22:12:08) [GCC 8.2.0] remote: ** Mercurial Distributed SCM (version 4.8.1+586-ef54bd33b476+20181224) remote: ** Extensions loaded: disable-lookup remote: Traceback (most recent call last): remote: File "/tmp/hgtests.i66Npc/install/bin/hg", line 43, in <module> remote: dispatch.run() remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/dispatch.py", line 99, in run remote: status = dispatch(req) remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/dispatch.py", line 225, in dispatch remote: ret = _runcatch(req) or 0 remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/dispatch.py", line 376, in _runcatch remote: return _callcatch(ui, _runcatchfunc) remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/dispatch.py", line 384, in _callcatch remote: return scmutil.callcatch(ui, func) remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/scmutil.py", line 166, in callcatch remote: return func() remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/dispatch.py", line 367, in _runcatchfunc remote: return _dispatch(req) remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/dispatch.py", line 1021, in _dispatch remote: cmdpats, cmdoptions) remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/dispatch.py", line 756, in runcommand remote: ret = _runcommand(ui, options, cmd, d) remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/dispatch.py", line 1030, in _runcommand remote: return cmdfunc() remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/dispatch.py", line 1018, in <lambda> remote: d = lambda: util.checksignature(func)(ui, *args, **strcmdopt) remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/util.py", line 1670, in check remote: return func(*args, **kwargs) remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/commands.py", line 5257, in serve remote: s.serve_forever() remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/wireprotoserver.py", line 797, in serve_forever remote: self.serveuntil(threading.Event()) remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/wireprotoserver.py", line 804, in serveuntil remote: _runsshserver(self._ui, self._repo, self._fin, self._fout, ev) remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/wireprotoserver.py", line 656, in _runsshserver remote: rsp = wireprotov1server.dispatch(repo, proto, request) remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/wireprotov1server.py", line 74, in dispatch remote: return func(repo, proto, *args) remote: File "/tmp/hgtests.i66Npc/install/lib/python/mercurial/wireprotov1server.py", line 195, in batch remote: data[k] = vals[k] remote: KeyError: 'namespace' abort: unexpected response: empty string [255] Differential Revision: https://phab.mercurial-scm.org/D5482

rust: core implementation for lazyancestors Once exposed through appropriate bindings, this should be able to replace ancestor.lazyancestors entirely. Differential Revision: https://phab.mercurial-scm.org/D5440

rust-cpython: binding for AncestorsIterator It's now reachable from Python as rustext.ancestor.AncestorsIterator Tests are provided in the previously introduced Python testcase: this is much more convenient that writing lengthy Rust code to call into Python. Differential Revision: https://phab.mercurial-scm.org/D5439

rust-cpython: implement Graph using C parents function We introduce the `Index` struct that wraps the C index. It is not intrinsically protected by the GIL (see the lengthy discussion in its docstring). Improving on this seems prematurate at this point. A pointer to the parents function is stored on the parsers C extension module as a capsule object. This is the recommended way to export a C API for consumption from other extensions. See also: https://docs.python.org/2.7/c-api/capsule.html In our case, we use it in cindex.rs, retrieving function pointer from the capsule and storing it within the CIndex struct, alongside with a pointer to the index. From there, the implementation is very close to the one from hg-direct-ffi. The naming convention for the capsule is inspired from the one in datetime: >>> import datetime >>> datetime.datetime_CAPI <capsule object "datetime.datetime_CAPI" at 0x7fb51201ecf0> although in datetime's case, the capsule points to a struct holding several type objects and methods. Differential Revision: https://phab.mercurial-scm.org/D5438

pull: fix inconsistent view of bookmarks during pull (issue4700) I had a share where a pull apparently pulled a bookmark but not the revision pointed to by the bookmark, which I suspect is due to this (and if not, we might as well remove known issues in this area). I do this by combining doing all the queries that could read the bookmarks in one round trip. I had to change the handling of the case where the server doesn't support the lookup query, because if it fails, it would otherwise make fremotebookmark.result() block forever. This is due to wireprotov1peer.peerexecutor.sendcommands's behavior (it fills a single future if any query fails synchronously and leaves all other futures unchanged), but I don't know if the fix is to cancel all other futures, or to keep going with the other queries. Differential Revision: https://phab.mercurial-scm.org/D5449

merge: modify the logical statement Differential Revision: https://phab.mercurial-scm.org/D5476

exthelper: correct a documentation typo

lfs: convert to using exthelper to wrap functions I'm not 100% sure that upgraderequirements() can be double annotated safely, but it seems OK based on printing the address of the function being wrapped. One thing I've noticed is that @eh.reposetup doesn't do the usual check to ensure that it's a local repo. Should that be baked into @eh.reposetup() somehow, possibly with a non-default option to skip the check? It seems like a gaping hole if every function that gets registered needs to add this check.

tests: convert a test extension to use exthelper This provides test coverage to uipopulate().

exthelper: drop fileset/revset/template support for now Yuya raised concerns about duplicating registrar functionality. There are a couple of ideas to work around this, which would allow bringing them back, and then backporting to evolve. For now, I just want to get the subsequent changes landed before the bulk b'' rewrite makes rebasing too hard.

exthelper: simplify configitem registration

extensions: import the exthelper class from evolve 980565468003 (API) This should help make extensions that wrap a lot of stuff more comprehendible. It was copied unmodified, except: - fix up the imports - rename final_xxxsetup() -> finalxxxsetup() to appease checkcode - avoid a [] default arg to wrapcommand() .. api:: Add `exthelper` class to simplify extension writing by allowing functions, commands, and configitems to be registered via annotations. The previous APIs are still available for use.

narrow: detect if narrowspec was changed in a different share With this commit, `hg share` should be usable with narrow repos. Design explained on https://www.mercurial-scm.org/wiki/NarrowSharePlan I was running into cache invalidation problems when updating the narrowspec. After spending a day trying to figure out a good solution, I resorted to just assigning repo.narrowpats and repo._narrowmatch after invalidating them. Differential Revision: https://phab.mercurial-scm.org/D5278

tests: add test for narrow+share For how narrow+share is supposed to work, see https://www.mercurial-scm.org/wiki/NarrowSharePlan. Differential Revision: https://phab.mercurial-scm.org/D5276

narrow: keep narrowspec backup in store As suggested by Yuya in review of D4099. Differential Revision: https://phab.mercurial-scm.org/D5470

tests: update narrowspec when narrowspec, not dirstate, is accessed test-narrow-expanddirstate.t mimics a Google-internal extension that updates the narrowspec whenever the dirstate is accessed. Since 1d09ba0d2ed3 (narrow: move remaining narrow-limited dirstate walks to core, 2018-10-01) and a few commits before it, we no longer restrict repo.dirstate.walk() to the narrowspec. It is instead done at a higher level (e.g. context.status()). We were running into problems with the Google-internal extension when importing those commits. The issue was that the narrowspec was read before the first dirstate access. I believe the right fix is to instead update the narrowspec when trying to read it (not when reading the dirstate), so that's what this patch does. Differential Revision: https://phab.mercurial-scm.org/D5275

merge: extract helper for creating empty "actions" dict Replicating the set of actions in multiple places is bad. Differential Revision: https://phab.mercurial-scm.org/D5472

manifest: accept narrowmatch into constructor instead of getting from repo The manifest should ideally not know at all about the repo, so this is just a little step towards cleaning that up. Differential Revision: https://phab.mercurial-scm.org/D5469

py3: byteify one more sys.argv in gendoc.py

test: introduce a new flag to display env variable line per line It's easier to conditionalize some of the environment variables per Mercurial version once there is only one value per line. Differential Revision: https://phab.mercurial-scm.org/D5453

revlog: add an explicit test for `issnapshot` We test the method on a real revlog containing "real" data.

revlog: add some direct testing of the slicing logic This test check slicing backed by an actual revlog. It will test the C version of slicing (if the test are run with the C extensions).

revlog: limit base to rev size ratio to 500 instead of 50 While a value of 50 provided large speedup in some case (400s → 7s) it also creates a slow down for a whole class of revision we are seeing in a private repository (0.1s → 3s). A value of 500 makes them disappear improving the total runtime (the slower revision still improve significantly (400s → 21s)). We need to run a wider array of tests on various repositories to see the effect on speed and size of different values for this acceptable constant. However, in the meantime, it seems safer to move back to a less impactful value.

contrib: remove unused version of os Caught by test-check-pyflakes.t Differential Revision: https://phab.mercurial-scm.org/D5471

py3: convert `'{}'.format(foo)` to `'%s' % foo` in the bookflow extension Byte strings don't have the former. Converting these to byte strings is waiting on the mass rewrite.

py3: byteify sys.argv in gendoc.py

repository: update interface signature of narrowmatch() This should have been part of 4fd0fac48922 (localrepo: allow narrowmatch() to accept matcher to intersect with, 2018-09-28) and 41fcdfe3bfeb (narrow: allow repo.narrowmatch(match) to include exact matches from "match", 2018-10-01). Differential Revision: https://phab.mercurial-scm.org/D5466

narrow: when narrowing, write new narrowspec before removing revlogs If revlogs were removed and then the process crashed before the narrowspec was written, the repo would be corrupt. Differential Revision: https://phab.mercurial-scm.org/D5467

narrow: replace "ui.warn(); return 1" by "raise error.Abort()" This is the usual way of doing it and I don't see a reason to do it differently here. Differential Revision: https://phab.mercurial-scm.org/D5468

test-bookmarks-pushpull: add failing test of issue4700 Differential Revision: https://phab.mercurial-scm.org/D5447

tests: update printenv.py argument parsing We are about to introduce a new flag for printing the HG environment variables one per line and it's easier to do when using the argparse module for argument parsing. Differential Revision: https://phab.mercurial-scm.org/D5452

fuzz: new fuzzer for parsers.fm1readmarkers Differential Revision: https://phab.mercurial-scm.org/D5465

server: always close http socket if responding with an error (issue6033) It's possible for hgweb to respond _very_ early with an error if we're catching certain types of errors. When we do, we need to tell the client the socket is toast when there's a POST involved because otherwise there can be lingering POST data on the socket that will confuse any future requests on the socket. This manifested as a flaky failure on Linux in an lfs extension test and a reliable failure on FreeBSD. With this patch applied, test-lfs-serve-access.t now passes for me on FreeBSD. Differential Revision: https://phab.mercurial-scm.org/D5498

match: fix assertion for fileset with no context (issue6046) A falsy changectx should be allowed.

templatekw: fix documentation typos

update: do not pass in user revspec as default destination (issue6044) When the revsingle() was introduced at 61c0df2b089a, it couldn't handle revspec=0 (not '0') properly. That's probably why the default was set to rev. This is technically BC since "hg update ''" was identical to "hg update '.'" whereas "hg update -r ''" is "hg update", but I believe that's a bug given no test fails with this change.

revlog: cache delta base value under -1 Such base are invalid so we better report them early.

revlog: catch revlog corruption in index_baserev A revision cannot use a base above itself, it can only happens one corrupted repository. Ignoring such corrupted could lead to infinite loop.

phabricator: properly encode boolean types in the request body I tripped over this playing with `hg debugcallconduit` to query for valid reviewers. If the JSON on stdin is written as 'True' or 'False', python complains it isn't valid JSON. If it's written as 'true' or 'false', it made it to the server, but got kicked back with this: abort: Conduit Error (ERR-CONDUIT-CORE): Error while reading "isBot": Expected boolean (true or false), got something else. The test isn't really relevant here (the code can be reverted, and it will pass), but this gives us coverage for the debug command.

parsers: better bounds checking in fm1readmarkers Our Python already calls this with reasonable values consistently, but my upcoming fuzzer is extremely quick to discover the lack of sanity checking here. Differential Revision: https://phab.mercurial-scm.org/D5464

fuzz: new fuzzer for dirstate parser Differential Revision: https://phab.mercurial-scm.org/D5463

fuzz: new fuzzer for revlog's parse_index2 method Differential Revision: https://phab.mercurial-scm.org/D5462

fuzz: extract Python initialization to utility package Avoids code duplication between fuzzers of parsers.so. Differential Revision: https://phab.mercurial-scm.org/D5461

fuzz: remove probably-wrong -fsanitize from fuzzutil.o rule Differential Revision: https://phab.mercurial-scm.org/D5460

parsers: remove long-dead parse_manifest method We haven't used this in years, I think it's fine to ditch it now. We had previously kept it around to ease bisecting with built extensions, but these days we've got a better versioning scheme anyway. Noticed this method kicking around while looking in parsers.so for likely fuzzing targets. Differential Revision: https://phab.mercurial-scm.org/D5459

help: hide default value for default-off flags If we no longer show the "[no-]" for default-off flags, it also seems unnecessary to show the "default: off" for them, since that's quite clearly the default. It's extra confusing for action flags like `hg bookmarks --delete`. Differential Revision: https://phab.mercurial-scm.org/D5455

help: show "[no-]" only for default-on Flags As Anton (av6) pointed out, the "[no-]" is confusing for action flags like `hg bookmark --delete`. We could come up with a way of indicating which flags are action flags (e.g. use None for the default value instead of False). However, it's probably also unlikely that users will want to negate even non-action flags like --hidden. One of the more common flags where the "[no-]" prefix would be useful is `hg evolve --update`. The reason it's helpful there is that it defaults to on. So I think we can simply include "[no-]" only for flags that are on by default (and thus require the user to add the "[no-]" for the option to have any effect). Note that there are use cases for negating flags that already off by default. For example, you may have an alias for `hg log -G --hidden -T foo` and now want to pass "--no-hidden" to that alias. However, I think that users who want that are likely to be advanced enough that they've already learnt about the "no-" prefix by seeing it somewhere else. Differential Revision: https://phab.mercurial-scm.org/D5454

shelve: drop unnecessary backup of narrowspec I mechanically added the backup code everywhere in ad24b581e4d9 (narrow: call narrowspec.{save,restore,clear}backup directly, 2018-08-03), but I can't think of a reason it would be needed in the shelve code, so let's drop it. Differential Revision: https://phab.mercurial-scm.org/D5457

shelve: pass transaction around to clarify where it's used Differential Revision: https://phab.mercurial-scm.org/D5456

py3: byteify docchecker The exception is printed as str because I'm too lazy to convert it and the pieces.

py3: replace hard coded `python` with $PYTHON in check-gendoc

py3: byteify gendoc.py This is mostly b'' prefixing, with some cargoculting of help.py to get around `textwrap.dedent()` and __doc__ string requirements.

py3: use bytes stdout in hghave.py This fixes a failure in test-run-tests.t around notarealhghavefeature. It seems crazy to me that all of this needs to be adjusted in all of these tests, but the line as run-tests.py sees it in _processoutput() before doing anything is already mangled with a trailing '\r'. Switching to normalizenewlines=True for TTest works, but I'm sure that breaks other stuff.

perfrevlogwrite: fix a typo in the option name The submitted patches was apparently damaged, and a 's' was missing.

fuzz: improve makefile per feedback from some folks at work This is a *ton* better. Once this lands, I can remove a nasty kludge from the oss-fuzz build.sh. Bonus: this fixes the coverage build. Differential Revision: https://phab.mercurial-scm.org/D5458

py3: use bytes stdout in test-check-help.t Setting stdout to binary seemed to have no effect on Windows, as it was appending a literal '\r' to each topic keyword. This also stops prepending 'b' to the topic on all platforms as well.

commandserver: preload repository in master server and reuse its file cache This greatly speeds up repository operation with lots of obsolete markers: $ ls -lh .hg/store/obsstore -rw-r--r-- 1 yuya yuya 21M Dec 2 17:55 .hg/store/obsstore $ time hg log -G -l10 --pager no (hg) 1.79s user 0.13s system 99% cpu 1.919 total (chg uncached) 0.00s user 0.01s system 0% cpu 1.328 total (chg cached) 0.00s user 0.00s system 3% cpu 0.180 total As you can see, the implementation of the preloader function is highly experimental. It works, but I'm yet to be sure how things can be organized. So I don't want to formalize the API at this point.

commandserver: add IPC channel to teach repository path on command finished The idea is to load recently-used repositories first in the master process, and fork(). The forked worker can reuse a warm repository if it's preloaded. There are a couple of ways of in-memory repository caching. They have pros and cons: a. "preload by master" pros: can use a single cache dict, maximizing cache hit rate cons: need to reload a repo in master process (because worker process dies per command) b. "prefork" pros: can cache a repo without reloading (as worker processes persist) cons: lower cache hit rate since each worker has to maintain its own cache c. "shared memory" (or separate key-value store server) pros: no need to reload a repo in master process, ideally cons: need to serialize objects to sharable form Since my primary goal is to get rid of the cost of loading obsstore without massive rewrites, (c) doesn't work. (b) isn't ideal since it would require much more SDRAMs than (a). So I take (a). The idea credits to Jun Wu.

upgrade: correct implementation of improvement.__ne__ The "not" operator binds more closely than "==": >>> not False == False False

extensions: use ui.log() interface to provide detailed loading information The output format changes and the messages will be sent to stderr instead of stdout, but I don't think that matters.

mq: implement log() on dummyui Otherwise ui.log() in extensions.py would explode.

ui: install logger that sends debug.extensions messages to stderr This will replace the custom log function introduced at d58958676b3c "extensions: add detailed loading information."

blackbox: resurrect recursion guard If I added ui.log() to hg.repository() function, test-merge-subrepos.t exploded. The problem is that the blackbox may create new repository instance while logging is active, and the created repository owns its new ui derived from the baseui, not from the ui which is processing the active logging. I tried to work around the issue in ui.log(), but that turned out to be not easy. We shouldn't globally lock the ui.log() since there may be more than one active repo/ui instances in threaded environment. We could store the logging state in thread-local storage, but that seems unnecessarily complex. So this patch reintroduces the _inlog flag to per-repository logger instances.

tests: filter out uninteresting log events This helps adding more log()s without updating the tests.

context: error out if basefilectx.cmp() is called without self._filenode The base implementation can't handle such cases because the filelog has no knowledge about the working directory. Loading self._filenode should have no extra cost since self.size() would load it anyway.

context: collapse complex condition to see if filelog have to be compared It's hard to read. I'd rather make the return statement duplicated.

merge with stable

worker: do not swallow exception occurred in main process Before, SystemExit(255) would be most likely raised since the worker processes were terminated by the main process and the status would be set to 255 in response. We should instead re-raise the exception occurred first. It's pretty hard to debug problems like the issue 6035 with no traceback.

run-tests: fix permission to clean up unreadable directories I found many hgtests.* directories left in $TMPDIR, which couldn't be deleted because test-ssh-repoerror.t created some directories with a-rx mode.

context: reimplement memfilectx.cmp() If I added a sanity check to basefilectx, test-context.py exploded. This patch copies the naive implementation from overlayworkingfilectx.

rust-cpython: build and support for Python3 Defined Cargo features for Python3, making them overall simpler to use, hooked them in build and made mercurial.rustext importable. This is tested with Python 3.6.7. Differential Revision: https://phab.mercurial-scm.org/D5446

py3: convert popen() command arguments in hgclient to str on Windows This fixes test-commandserver.t and test-keyword.t, which was previously complaining TypeError("a bytes-like object is required, not 'str'")

py3: byteify tests/pullext.py # skip-blame for b'' prepending

py3: byteify the fakeversion extension in test-http-bad-server.t # skip-blame for b'' prefixing

py3: spawn all python instances with legacy stdio enabled on Windows This fixes 6 more tests. See ef7119cd4965.

setup: avoid attempting to invoke the system-wide hg.exe on Windows On Windows, the executable in the current directory gets priority over anything in $PATH (both for cmd.exe and MSYS). That means, the former code was launching the local hg.exe instead of the system-wide one, if it was previously built. If that failed, it then fell back to the local hg code, but run through python.exe. I'm not sure what it is about ef7119cd4965, but that started throwing up a messagebox that python37.dll couldn't be loaded. (And indeed, python37 is not in $PATH by default.) Invoking the local hg via the current python avoids that.

delta: ignore base whose chains already don't match expectations If we know the existing chain does not match our criteria, there is no point to build a delta to append. This is especially useful when dealing with a full text much smaller than its parent. In that case, the parent chain is probably already too large. example affected manifest write before: 1.421005s after: 0.815520s (-42%)

delta: exclude base candidate much smaller than the target If a revision's full text is that much bigger than a base candidate full text, we no longer consider that candidate. This solves a pathological case we encountered on a very specify repository. It contains a long series of changesets with a very small manifest (one file) co-existing with others changesets using a very large manifest. Without this filtering, we ended up considering a large number of tiny full snapshots as a potential base. It resulted in very large delta (the size of the full text) and mercurial spending 99% of its time compressing these deltas. The timing of a commit moved from about 400s to about 10s (still slow, but not ridiculously slow).

perfrevflogwrite: clear revlog cache between each write We want to measure write time from a cold cache (similar to commit). So we need to clear the cache to prevent computation from rev N-1 to interfere with rev N.

py3: enable legacy stdio mode in exewrapper This drops the test failure count from 166 to 117. The failures were typically in the form of `hg serve -d` spawning but crashing immediately, leaving clients with "bad http status" lines, connection refusals, and so forth. The underlying message on the server side was either "OSError: [WinError 6] The handle is invalid" or "OSError: [WinError 1] Incorrect function". Additionally, no output was rendered if the pager was activated. Thanks to Yuya for diagnosing the problem. The failure count drops to 107 when PYTHONLEGACYWINDOWSSTDIO=1 is defined in the environment. These failures seem to revolve around the dummyssh server process, and dumbhttp.py. So I'll probably add that to the test runner. One small regression here (only in py3) is that if hg.exe is already built, a messagebox appears when building it again saying that python37.dll can't be loaded. Python3 isn't in PATH by default, and setup.py tries running bare `hg` first. But MSYS prepends '.' to PATH, so it runs the local hg, but can't find the library. It falls back to the python used to invoke setup.py, so ultimately it works. I'm not sure if it's better to strip '.' from PATH or just skip right to `sys.executable hg` on Windows. Also, something seems to be wrong with run-tests._usecorrectpython(). I accidentially left off the 'PYTHON="py -3"' when building (thus making py2 stuff), and yet managed to invoke run-tests.py with "py -3". (And that only had 67 failures.)

run-tests: alias hg to hg.exe on Windows To enable legacy stdio mode on Windows, hg.exe needs to be updated. But before that, we actually have to use it when running the tests. I *think* what was happening before was when MSYS invoked `hg`, it looked at the shbang line and ran python.exe found there. The test harness must be updating $PATH to include the python used to launch it, and therefore it ran py3. As a side note, this also fixed `py -3 run-tests.py` (without --local), which complained about the space in the shbang line before this. This should also help in WSL, because the explicit '.exe' is needed to invoke a Windows app instead of invoking the Linux app. I have no idea why this change capitalizes Lib in the tests, as it was previously lowercase for both py2 and py3.

py3: conditionalize the python version in test-install.t

tests: followup on network related errors on Debian 9 First, update test-clonebundles.t which was failing since the introduction of the `$EADDRNOTAVAIL$` common pattern. Also update two tests from 8695fbe17f7c with a more compact condition line. Differential Revision: https://phab.mercurial-scm.org/D5445

py3: unimplement RevlogError.__str__() On Python 2, str(exc) would crash if __str__() returned a unicode containing non-ASCII characters.

py3: use forcebytestr() to stringify hook exception This fixes test-narrow-exchange.t.

color: fix a documentation typo

py3: quote $PYTHON in test-patchbomb.t for Windows I couldn't get the quoting right in the environment variable, so now it's a function.

rust-cpython: testing the bindings from Python This is easier and more convincing than doing the same tests from a Rust tests module. Differential Revision: https://phab.mercurial-scm.org/D5437

rust-cpython: build via HGWITHRUSTEXT=cpython The existing behaviour, building the direct ffi bindings if HGIWTHRUSTEXT is just set is unchanged, but if HGWITHRUSTEXT is cpython, then the cpython bindings (aka mercurial/rustext.so) are built. Differential Revision: https://phab.mercurial-scm.org/D5436

rust: better treatment of cargo/rustc errors Differential Revision: https://phab.mercurial-scm.org/D5435

rust-cpython: start cpython crate bindings This changeset introduces the hg-cpython crate, that compiles as a shared library holding a whole Python package (mercurial.rustext), with only the empty 'ancestor' submodule for now. Such bindings will be easier and safer to develop and maintain that those of `hg-direct-ffi`. They don't involve C code, only unsafe Rust that's mostly isolated within the cpython crate. The long-term goal would be to import the provided modules, such as rustext.ancestor with mercurial.policy.importmod, same as we already do with cext modules. Differential Revision: https://phab.mercurial-scm.org/D5434

rust-cpython: exclude hgcli from workspace hgcli uses a specific rust-cpython commit by indygreg, of which a PR has been derived which is not merged nor released yet. But we can't use several versions of the sys-python2.7 crate in a single workspace: it makes for a build error. Since hgcli does not at the time being need anything from hg-core, whereas the upcoming hg-cpython will. So for now we're moving hgcli aside, hoping we could base all of them on the same version of rust-cpython again in the future. Differential Revision: https://phab.mercurial-scm.org/D5433

sparse-revlog: protect C code against delta chain including nullrev For unclear reasons, some repositories include nullrev (-1). Re-computing delta for such repo remove nullrev from all chain, so some older versions have been creating them. This currently raise an IndexError with the new C code doing chain slicing as it expect all item to be positive. Both python and C code for reading delta chain preserve nullrev, and the Python code for chain slicing handle the case fine. So we take the safe route and make the new C code works fine in that case.

sparse-revlog: handle nullrev in index_get_length The more generic index_get method handle nullrev fine, we apply the same logic here.

sparse-revlog: handle nullrev in index_get_start The more generic index_get method handle nullrev fine, we apply the same logic here.

revlog: introduce a constant for nullrev in `revlog.c` The value is important enough to be explicitly tracked.

rust: translation of missingancestors This is as direct as possible a translation of the ancestor.missingancestors Python class in pure Rust. The goal for this changeset is to make it easy to compare with the Python version. We also add to Python tests the cases that helped us develop and debug this implementation. Some possible optimizations are marked along the way as TODO comments Differential Revision: https://phab.mercurial-scm.org/D5416

contrib: provide a small script that draw performance plot We have been using this script to look into the result of various runs of the `hg perfrevlogwrite` command. It seems useful enough to be shared more widely.

delta: filter nullrev out first When picking a potential candidate, we filter them on various criteria. The "different from nullrev" criteria is very fast to compute and we should process it first.

perf: report more of the higher range in perfrevlogwrite Since the delta chain length is limited to 1000 revisions, we get a new snapshot about every 1000 revisions. If we assume that the snapshot will be most of the slowest revision, the current display (99% and max) are not very precise in their area. We now include more information about this space in the default report.

help: present boolean arguments as "--[no-]foo" This should make it much more discoverable (we document it in `hg help flags`, but most users don't think to look there). Note that flags that default to None (and not False) will not get this new presentation. We can change the defaults to False later for flags where it makes sense (probably almost all boolean flags). Differential Revision: https://phab.mercurial-scm.org/D5432

help: show "default: off" for boolean flags that default to off Differential Revision: https://phab.mercurial-scm.org/D5431

help: use "default: on" instead of "default: True" "True" feels like a Python thing and not something that users should see. Differential Revision: https://phab.mercurial-scm.org/D5430

check-commit: disallow capitalization only right after topic It should be okay to write "topic: change 'default: True' to 'default: on'". Differential Revision: https://phab.mercurial-scm.org/D5429

absorb: don't prompt to apply changes when there are none to apply Differential Revision: https://phab.mercurial-scm.org/D5428

py3: add a b'' prefix in tests/test-remotefilelog-pull-noshallow.t This makes the test pass on Python 3. # skip-blame because just b'' prefix. Differential Revision: https://phab.mercurial-scm.org/D5427

error: implement __str__ on RevlogError to fix some output defects on Py3 We open-code encoding.unimethod here to avoid cycles, and do a local import of encoding when someone str()s a RevlogError. It's not my favorite solution, but it gets the job done. Differential Revision: https://phab.mercurial-scm.org/D5426

perf: add two more missing b prefixes for Python 3 # skip-blame just two more b prefixes Differential Revision: https://phab.mercurial-scm.org/D5425

match: fix doctest to use bytes instead of str Fixes doctests on Python 3. Differential Revision: https://phab.mercurial-scm.org/D5423

py3: use pycompat.bytechr() instead of chr() in test-fastannotate-revmap.py This makes the test pass on Python 3. Differential Revision: https://phab.mercurial-scm.org/D5422

py3: add a missing b'' prefix in contrib/perf.py # skip-blame because just b'' prefixes This fixes test-contrib-perf.t on Python 3 which started failing. Differential Revision: https://phab.mercurial-scm.org/D5421

py3: use '%d' for integers instead of '%s' This should fix test-rebase-inmemory.t which started failing on Python 3 after recent changes. Differential Revision: https://phab.mercurial-scm.org/D5420

py3: whitelist another passing test caught by buildbot Differential Revision: https://phab.mercurial-scm.org/D5419

py3: fix bad escapes of sub() replacement pattern in test-template-basic.t Python 3.7 starts complaining about it. We have to double the backslash or '\x5c' to get around.

perfdiscovery: benching findcommonheads() This works between the local repo and any peer given by its path, and should be useful for further work on discovery Differential Revision: https://phab.mercurial-scm.org/D5418

windows: ensure pure posixfile fd doesn't escape by entering context manager There are tests in test-revlog-mmapindex.t and test-rebase-mq-skip.t that are fixed by this, but we usually don't use --pure on Windows. For whatever reason, the remaining --pure failures are various errors like $ENOTDIR$ and "Access is denied" have a trailing '.'.

vfs: ensure closewrapbase fh doesn't escape by entering context manager I'm not sure if there's a problem in practice here, as there's no test failure either way. The __exit__() and close() methods raise an exception, so maybe __exit__() and close() are being called directly on the underlying handle when delayclosedfile is used on a context manager? I doubt that was intended.

windows: ensure mixedfilemodewrapper fd doesn't escape by entering context mgr Otherwise it seems that the special read and write handling would be bypassed.

py3: ensure the proxied Windows fd doesn't escape by entering context manager The purpose of the proxy class is to provide the `name` attribute which contains the file path. But in tests that used a context manager, it still blew up complaining that 'int' doesn't have a 'startswith' function.

merge with stable

templatefuncs: specialize "no match" value of search() to allow % operation If Python had Maybe or Option, the type of the search() result would be Option<Mapping>, which can be considered as a 0/1 container of a Mapping. So it makes sense that {search(r'no match pattern', x) % "whatever"} is mapped to an empty string.

templatefuncs: add regexp search() function that extracts substring This can be used to extract an issue number from a commit message, for example: {search(r'\(issue([0-9]*)\)', desc) % '{1}'}

rust: changed Graph.parents to return [Revision; 2] This will allow for simple iteration on parent revisions, such as: for parent in graph.parents(rev)?.iter().cloned() This seems to be a zero overhead abstraction once built in release mode. Differential Revision: https://phab.mercurial-scm.org/D5415

rust: improved docstring In the previous wording, rustfmt wanted to cut at the == which is not very readable in my taste. Differential Revision: https://phab.mercurial-scm.org/D5414

revset: move subscript relation functions to its own dict This will help adding more relation functions in extensions. We skip short names (that consist of one letter) while raising UnknownIdentifier because such names cannot be suggested anyway: the similarity cutoff in dispatch._getsimilar() is currently 0.6.

py3: teach run-tests.py to handle exe with spaces when --local isn't specified This was the reason that no amount of quoting worked in test-hghave.t. `os.popen()` needed to be swapped out because while the added quoting around line 3124 worked on py3, it failed on py2. See 38d51371792b. The problem with `os.system()` was wrongly splitting the command on the space in 'Program Files', regardless of quoting. It looks like there are a few other instances of `os.system()` in core code, so presumably those should be replaced?

rust: adapted hg-core tests for iteration over Result Now AncestorsIterator iters on Result<Revision, GraphError>

win32: close the handles associated with a spawned child process Probably not a big deal because at this point, the call is only used when spawning a daemonized server. In that case, the parent process goes away first, so it won't prevent the child from being cleaned up.

rust: remove comment about error handling of AncestorsIterator To be align with 443eb4bc41af "rust: propagate error of index_get_parents() properly." Spotted by Georges Racinet.

test: fix test-http-bad-server with current python 2.7 https://github.com/python/cpython/pull/2825 changed the exception message for empty http status line. Differential Revision: https://phab.mercurial-scm.org/D5412

perf: add perfprogress command I've noticed that progress bars can add significant overhead to tight loops. Let's add a perf command that attempts to isolate that overhead. With a default hgrc, iteration over 1M items appears to take ~3.75s on my machine. Profiling reveals ~28% of time is spent in ui.configbool() resolving the value of the progress.debug config option. Even if I set progress.disable=true, execution still takes ~2.60s, with ~59% of the time spent in ui.configbool(). Differential Revision: https://phab.mercurial-scm.org/D5407

wireprotov2: unify file revision collection and linknode derivation The old mechanism for choosing which file revisions to send in the haveparents=True case was buggy in multiple ways - the most severe of which being that file revisions were excluded when they shouldn't have been. This commit unifies the logic for deriving the filenodes that will be sent by the "filesdata" command. We now consistently read files data from manifests. The "haveparents" argument now controls whether we iterate ctx.files() or use the full manifest to derive relevant files. The logic here is still woefully lacking to fully support shallow clones. It will require an API break to fully address. This commit should at least make the server APIs emit proper data, which is strictly better than before. Differential Revision: https://phab.mercurial-scm.org/D5406

wireprotov2: send linknodes to emitfilerevisions() Previously, linknodes were calculated within emitfilerevisions() by using filectx.introrev(), which would always use the linkrev/linknode as recorded by storage. This is wrong for cases where the receiver doesn't have the changeset the linknode refers to. This commit changes the logic for linknode emission so the mapping of filenode to linknode is computed by the caller and passed into emitfilerevisions(). As part of the change, linknodes for "filesdata" in the haveparents=False case are now correct: the existing code performed a manifest walk and it was trivial to plug in the correct linknode. However, behavior for the haveparents=True case is still wrong because it relies on filtering linkrevs against the outgoing set in order to determine what to send. This will be fixed in a subsequent commit. The change test test-wireproto-exchangev2-shallow.t is a bit wonky. The test repo has 6 revisions. The changed test is performing a shallow clone with depth=1. So, only file data for revision 5 is present locally. So, the new behavior of associating the linknode with revision 5 for every file revision seems correct. Of course, when backfilling old revisions, we'll want to update the linknode. But this problem requires wire protocol support and we'll cross that bridge later. Differential Revision: https://phab.mercurial-scm.org/D5405

tests: add tests for server-side linknode adjustment with wireprotov2 The current implementation of linknode serving in wireprotov2 simply serves up the linkrev/linknode as stored: it doesn't attempt to adjust the linknode to what the receiver is aware of. This can result in the client seeing a linknode referencing a changeset that is unknown to it. This commit adds test coverage of that scenario. The tests in test-wireproto-command-filesdata.t demonstrate two failures. First, the linknode refers to a changeset not in the available set. Second, the server doesn't send a file revision that it should have (because of linkrev filtering). The test in test-wireproto-exchange.t demonstrates that the lack of a file revision results in a corrupted repository on the client. Differential Revision: https://phab.mercurial-scm.org/D5404

tests: fix empty commit in test This was almost certainly a bug. Differential Revision: https://phab.mercurial-scm.org/D5403

tests: add sparserevlog requirement to test-sqlitestore.t Looks like this test was missed when enabling sparse revlogs by default. Differential Revision: https://phab.mercurial-scm.org/D5402

tests: remove all transitional configuration Now that sparse-revlog is enabled by default, we no longer needs it. Differential Revision: https://phab.mercurial-scm.org/D5346

sparse-revlog: enabled by default The feature provides large benefits. It now seems mature enough to be enabled by default. * It solves catastrophic issues regarding delta storage in revlog, * It allows for shorter delta chain in all repositories, improving performances. Running benchmark of a wide range of operation did not reveal problematic impact. Performance gains are observed where expected. The format is supported by Mercurial version 4.7. So it seems safe to enable it by default now. Here is a reminder of key numbers regarding this delta strategy effect on repository size and performance. Effect on Size: =============== For repositories with a lot of branches, sparse-revlog significantly improve size, fixing limitation associated with the span of a delta chain. In addition, sparse-revlog, deal well with limitations of the delta chain length. For large repositories, this allows for a stiff reduction of the delta chain without a problematic impact on the repository size. This delta chain length improvement helps all repositories, not just the ones with many branches. As a reminder, here are the default chain limits for each "format": * no-sparse: none * sparse: 1000 Mercurial --------- Manifest Size: limit | none | 1000 ------------|-------------|------------ no-sparse | 6 143 044 | 6 269 496 sparse | 5 798 796 | 5 827 025 Manifest Chain length data limit || none || 1000 value || average | max || average | max ------------||---------|---------||---------|--------- no-sparse || 429 | 1 397 || 397 | 1 000 sparse || 326 | 1 290 || 313 | 1 000 Full Store Size limit | none | 1000 ------------|-------------|------------ no-sparse | 46 944 775 | 47 166 129 sparse | 46 622 445 | 46 723 774 pypy ---- Manifest Size: limit | none | 1000 ------------|-------------|------------ no-sparse | 52 941 760 | 56 200 970 sparse | 26 348 229 | 27 384 133 Manifest Chain length data limit || none || 1000 value || average | max || average | max ------------||---------|---------||---------|--------- no-sparse || 769 | 3 889 || 390 | 1 000 sparse || 1 223 | 3 846 || 495 | 1 000 Full Store Size limit | none | 1000 ------------|-------------|------------ no-sparse | 336 050 203 | 339 309 413 sparse | 338 673 985 | 339 709 889 Mozilla ------- Manifest Size: limit | none | 1000 ------------|----------------|--------------- no-sparse | 215 096 339 | 1 708 853 525 sparse | 188 947 271 | 278 894 170 Manifest Chain length data limit || none || 1000 value || average | max || average | max ------------||---------|---------||---------|-------- no-sparse || 20 454 | 59 562 || 491 | 1 000 sparse || 23 509 | 69 891 || 489 | 1 000 Full Store Size limit | none | 1000 ------------|----------------|--------------- no-sparse | 2 377 578 715 | 3 876 258 798 sparse | 2 441 677 137 | 2 535 997 381 Netbeans -------- Manifest Size: limit | none | 1000 ------------|----------------|--------------- no-sparse | 130 088 982 | 741 590 565 sparse | 118 836 887 | 159 161 207 Manifest Chain length data limit || none || 1000 value || average | max || average | max ------------||---------|---------||---------|--------- no-sparse || 19 321 | 61 397 || 510 | 1 000 sparse || 21 240 | 61 583 || 503 | 1 000 Full Store Size limit | none | 1000 ------------|----------------|--------------- no-sparse | 1 160 013 008 | 1 771 514 591 sparse | 1 164 959 988 | 1 205 284 308 Private repo #1 --------------- Manifest Size: limit | none | 1000 ------------|-----------------|--------------- no-sparse | 33 725 285 081 | 33 724 834 190 sparse | 350 542 420 | 423 470 579 Manifest Chain length data limit || none || 1000 value || average | max || average | max ------------||---------|---------||---------|--------- no-sparse || 282 | 8 885 || 113 | 1 000 snapshot || 3 655 | 8 951 || 530 | 1 000 Full Store Size limit | none | 1000 ------------|----------------|--------------- no-sparse | 41 544 149 652 | 41 543 698 761 sparse | 8 448 037 300 | 8 520 965 459 Effect on speed: ================ Performances are strongly impacted by the delta chain length. Longer chain results in slower revision restoration. For this reason, the 1000 chain limit introduced by sparse-revlog helps repository with previously large chains a lot. In our corpus, this means `netbeans` and `mozilla-central` who suffered from unreasonable manifest delta chain length. Another way sparse revlog helps, is by producing better delta's. For repositories with many branches, the pathological patterns that resulted in many sub-optimal deltas are gone. Smaller delta help with operations where deltas are directly relevant, like bundle. However, the sparse-revlog logic introduces some extra processing and a more throughout testing of possible delta candidates. Adding an extra cost in some cases. This cost is usually counterbalanced by the other performance gain. However, for smaller repositories not affected by delta chain length issues or branching related issues, this might make things a bit slower. However, these are also repository where revlog performance is dwarfed by other costs. Below are the summary of some timing from the performance test suite running at `http://perf.octobus.net/` for a handful of key commands and operation. It is important to keep in mind that most of this command works on the tip part of the repository. The non-sparse and sparse version produce different delta chains and the tip revision can end up at an arbitrary point of these chains. This will impact some performance number listed in this summary. For the record: here is the delta chain length for the tip revision of manifest log in the benchmarked repository: | no-sparse | sparse | mercurial | 94 | 904 | pypy | 23 | 673 | netbeans | 4158 | 258 | mozilla | 63263 | 781 | As you can see, the chain length for mercurial and pypy turn out to be significantly longer. The netbeans and mozilla one get shorter because these repositories benefit from the maximum chain length. Timing for `hg commit`: ----------------------- The time taken by `hg commit` does not varies significantly, no drawback for using sparse here. | no-sparse | sparse | mercurial | 68.1ms | 66.7ms | pypy | 95.0ms | 94.1ms | netbeans | 614.0ms | 611.0ms | mozilla | 1340.0ms | 1.320.0ms | Check the final section for statistics on a wider array of write. Timing for bundling 10 000 changesets ------------------------------------- The repository that benefits from better delta see a good performance boost. The other ones are not significantly affected. | no-sparse | sparse | mercurial | 3.1s | 3.0s | pypy | 25.1s | 7.5s | netbeans | 24.2s | 17.0s | mozilla | 23.7s | 25.0s | Timing for unbundling 1 000 changesets -------------------------------------- Mercurial and mozilla are unaffected. The pypy repository benefit well from the better delta. However, the netbeans repository takes a visible hit. Digging that difference reveals that it comes from the sparse-revlog bundle having to deal with a snapshot that was re-encoded in the bundle. The slow path for adding new a revision had to be triggered for it, slowing things down. The Sparse versions do not have such snapshot to handle similar cases in the tested configuration. | no-sparse | sparse | mercurial | 519ms | 502ms | pypy | 1.270ms | 886ms | netbeans | 1.370ms | 2.250ms | mozilla | 3.230ms | 3.210ms | Netbeans benefits from the better deltas in other dimensions too. For example, the produced bundle is significantly smaller: * netbeans-no-sparse.hg: 2.3MB * netbeans-sparse.hg: 1.9MB Timing to restore the tip most manifest entry: ---------------------------------------------- Nothing surprising here. The timing for mercurial and pypy are within a small range where they won't affect performance much. In our tested case, they are slower as they use a longer chain. Timing for netbeans and mozilla improves a lot. Removing a significant amount of time. | no-sparse | sparse | mercurial | 1.09ms | 3.15ms | pypy | 4.11ms | 10.70ms | netbeans | 239.00ms | 112.00ms | mozilla | 688.00ms | 198.00ms | Reading 100 revision in descending order: ----------------------------------------- We see the same kind of effect when reading the last 100 revisions. Large boost for netbeans and mozilla, as they use much smaller delta chain. Mercurial and pypy longer chain means slower reads, but nothing gets out of control. | no-sparse | sparse | mercurial | 0.089s | 0.268s | pypy | 0.259s | 0.698s | netbeans | 125.000s | 20.600s | mozilla | 23.000s | 11.400s | Writing from full text: statistic for the last 30K revisions ------------------------------------------------------------ This benchmark adds revisions to revlog from their full text. This is similar to the work done during a commit, but for a large amount of revisions so that we get a more relevant view. We see better overall performances with sparse-revlog. The very worst case is usually slower with sparse-revlog, but does not gets out of control. For the vast majorities of the other writes, sparse-revlog is significantly faster for larger repositories. This is reflected in the accumulated rewrite time for netbeans and mozilla. The notable exception is the pypy repository where things get slower. The extra processing is not balanced by shorter delta chain. However, this is to be seen as a blocking issue. First, the overall time spend dealing with revlog for the repository pypy size is small compared to the other costs, so we get slower on operations that matter less than for other larger repository. Second, we still get nice size benefit from using sparse-revlog, smaller repo size brings other usability and speed benefit (eg: bundle size). max time | no-sparse | sparse | mercurial | 0.010143s | 0.045280s | pypy | 0.034924s | 0.243288s | netbeans | 0.605371s | 2.130876s | mozilla | 1.478342s | 3.424541s | 99% time | no-sparse | sparse | mercurial | 0.003774s | 0.003758s | pypy | 0.017387s | 0.025310s | netbeans | 0.576913s | 0.271195s | mozilla | 1.478342s | 0.449661s | 95% time | no-sparse | sparse | mercurial | 0.002069s | 0.002120s | pypy | 0.010141s | 0.014797s | netbeans | 0.540202s | 0.258644s | mozilla | 0.654830s | 0.243440s | full time | no-sparse | sparse | mercurial | 14.15s | 14.87s | pypy | 90.50s | 137.12s | netbeans | 6401.06s | 3411.14s | mozilla | 3086.89s | 1991.97s | Differential Revision: https://phab.mercurial-scm.org/D5345

test: preemptively disable sparse-revlog for some of test-upgrade-repo Some part of this test assumes sparse-revlog is off. We enforce it explicitly. Differential Revision: https://phab.mercurial-scm.org/D5344

test: enable sparse-revlog for test-wireproto-content-redirects.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5343

test: enable sparse-revlog for test-wireproto-command-capabilities.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5342

test: enable sparse-revlog for test-wireproto-caching.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5341

test: enable sparse-revlog for test-treemanifest.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5340

test: enable sparse-revlog for test-stream-bundle-v2.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5339

test: enable sparse-revlog for test-ssh.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5338

test: enable sparse-revlog for test-ssh-proto-unbundle.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5337

test: enable sparse-revlog for test-ssh-proto.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5336

test: enable sparse-revlog for test-ssh-bundle1.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5335

test: enable sparse-revlog for test-sparse-requirement.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5334

test: enable sparse-revlog for test-revlog-v2.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5333

test: enable sparse-revlog for test-repo-compengines.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5332

test: enable sparse-revlog for test-remotefilelog-log.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5331

test: enable sparse-revlog for test-remotefilelog-clone-tree.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5330

test: enable sparse-revlog for test-remotefilelog-clone.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5329

test: enable sparse-revlog for test-phases.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5328

test: enable sparse-revlog for test-narrow-clone.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5327

test: enable sparse-revlog for test-narrow-clone-stream.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5326

test: enable sparse-revlog for test-narrow-clone-no-ellipsis.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5325

test: enable sparse-revlog for test-lfs-serve.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5324

test: enable sparse-revlog for test-lfs-largefiles.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5323

test: enable sparse-revlog for test-lfconvert.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5322

test: enable sparse-revlog for test-init.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5321

test: enable sparse-revlog for test-http-protocol.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5320

test: enable sparse-revlog for test-hgweb-commands.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5319

test: enable sparse-revlog for test-debugcommands.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5318

test: enable sparse-revlog for test-clone-uncompressed.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5317

test: enable sparse-revlog for test-clonebundles.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5316

test: enable sparse-revlog for test-bundle.t We are about to enable sparse-revlog globally. To help with reviewing the tests change, we isolate them in individual changesets. Differential Revision: https://phab.mercurial-scm.org/D5315

test: make sure sparse-revlog does not interfer with test-parseindex.t That test carefully craft revlog to reproduce issues. We do not want new revlog behavior to interfere with that. Differential Revision: https://phab.mercurial-scm.org/D5314

test: explicitly disable sparse-revlog in test-http-bad-server.t The `test-http-bad-server.t` test is very fragile with hard-coded number all around. Adding any new capability have massive consequences on it that are long and tiresome to handle. Since the test check more generic protocol level behavior, it does not need to follow all new capabilities. Disabling the capabilities is a much simpler way to avoid the issue. We make sure `sparse-revlog` will not impact this test once turned on by default. Differential Revision: https://phab.mercurial-scm.org/D5313

test: explicitly disable sparse-revlog in test-generaldelta.t That test is about testing generaldelta without further improvement. We ensure it will stay that way even if sparse-revlog is turned on by default. The sparse-revlog format is tested on his own in a different test file. Differential Revision: https://phab.mercurial-scm.org/D5312

sparse-revlog: disable sparse-revlog if config disable general-delta Sparse-revlog requires general-delta in order to work. If general-delta is explicitly disabled through configuration, we should honor that and also disabled sparse-revlog. This change will be more relevant when sparse-revlog is enabled by default. Differential Revision: https://phab.mercurial-scm.org/D5311

test: fix config typo in test-upgrade-repo.t Differential Revision: https://phab.mercurial-scm.org/D5310

tests: use $RUNTESTDIR instead of $TESTDIR in narrow-library.sh This is done so that third party extensions can use narrow-library.sh in tests. Differential Revision: https://phab.mercurial-scm.org/D5400

tests: update network related errors for Debian 9 We have a CI job that runs the Mercurial tests in parallel. Some of the network related failures seems to be different on the environment. Oddly, those failures happens only when running the tests in parallel, not when running the test file only. I have no idea how to get the windows formatted message for the error, if someone could give me an hand, I will update this changeset with the value. Differential Revision: https://phab.mercurial-scm.org/D5401

util: implement pop() on lrucachedict This moves __delitem__() to pop() as the requirement is pretty much the same, and reimplement __delitem__() by using pop().

util: add method to peek item in lrucachedict I want a function that doesn't unnecessarily update the internal state of the cache dict after fork().

commandserver: loop over selector events An IPC socket will be waited by the same selector.

commandserver: remove redundant "if True" block

commandserver: extract handler of new socket connection This prepares for adding an IPC socket.

commandserver: pass around option to hook repo instance creation This is necessary to wrap a repo instance so the master process will be notified on repo.close().

py3: stop subscripting socket.error In 3.3 and later, this is now an alias for OSError. I hacked up the server code enough that I was able to trigger the exception handler in server.py from test-http-bundle1.t. Other instances of this either subscript through the `args` member, or reference the errno or strerror attributes. Note that on Windows, the errno value seems to reflect the Winsock error, so the various tests for EPIPE seem like they would always fail. But that seems to be the case in py2 as well.

test: update test-logtoprocess.t to not requires less The test previously assumed that a working pager was present in the test environment. Update it to use the fakepager instead. Differential Revision: https://phab.mercurial-scm.org/D5399

py3: quote several instances of $PYTHON for Windows Python3 lives in Program Files by default. The last remaining unquoted instance is in test-hghave.t. I can't figure out how to quote that, as it either complains: 'c:\\Program' is not recognized as an internal or external command, operable program or batch file. or $TESTTMP.sh: line 22: "C:/Program Files/Python37/python.exe": $ENOENT$

py3: adapt test-largefiles-wireproto.t to python3

py3: byteify test-push-race.t

py3: convert filename to bytes in test-hardlinks.t Windows was complaining about passing str to CreateFileA() in win32.py.

narrow: keep bookmarks temporarily stripped for as long as commits are The narrow extension also has support for shallowness and for inserting older commits on pull. It works by temporarily stripping newer commits, adding the older commits, then re-applying the stripped bundle. The regular Mercurial server uses that when you widen, although it shouldn't be necessary there. Our Google-internal server does it when the user requests an older commit. Our Google-internal tests fail since 7caf632e30c3 (filecache: unimplement __set__() and __delete__() (API), 2018-10-20). I haven't quite understood the problem, but it's related to the way we temporarily hide bookmarks while the commits they point to are stripped. When a transaction is started, Mercurial tries to read various things from the repo for the transaction summary. That leads to computation of hidden commits, which leads to an attempt to find commits pinned by bookmarks. This is the reason we temporarily hide the bookmarks. I think the aforementioned commit makes the restored bookmarks visible earlier than before (which seems like an improvement), so we can no longer incorrectly rely on the repo._bookmarks field being cached too long (IIUC). This patch makes it so we restore the temporarily hidden bookmarks only after the temporary bundle has been re-applied. It also adapts the code to update the repo.__bookmarks field using the pattern described in the aforementioned commit instead of writing directly to the fiels. I have spent many hours trying to understand what was going on here, but I still don't know if this can also happen without our custom server. So this patch unfortunately does not add any tests; I have only been able to test the fix using our Google-internal tests. Differential Revision: https://phab.mercurial-scm.org/D5398

narrow: drop obsolete support for old Mercurial Not needed since narrow is shipped with core. Differential Revision: https://phab.mercurial-scm.org/D5397

tests: test that narrow preserves bookmarks on widen When widening inserts older commits in the changelog, we have to preserve bookmarks so they are not removed by the call to repair.strip() we temporarily do. We didn't have any tests for that before. Differential Revision: https://phab.mercurial-scm.org/D5396

rebase: clarify that commits that become empty are skipped Our message just said that it created no changes, but we didn't explicitly say that that meant that the the commit was not rebased. It also wasn't clear why it created no changes, so this patch also clarifies that that was because the changes were already in the destination. Differential Revision: https://phab.mercurial-scm.org/D5395

rebase: format commit in usual way when explaining that it became empty We use the _ctxdesc() for other similar messages, so it seems we should use it here too. Differential Revision: https://phab.mercurial-scm.org/D5394

rust: propagate error of index_get_parents() properly Before, rustla_contains() would return 0 on error, and the exception would be cleared or noticed somewhere else. We need to propagate the error from AncestorsIterator up to the FFI surface.

rust: look up HgRevlogIndex_GetParents() from symbol table And removes the unused index_get_parents_checked() function. I expect the Index struct will be turned into a pyobject type, though I haven't written any PoC-level patches yet.

revlog: add public CPython function to get parent revisions Since this is a public function, it validates the input revision, and supports nullrev. index_get_parents_checked() will be replaced by this function.

revlog: rename indexType to HgRevlogIndex_Type as it's a global symbol It follows the CPython naming convention, but uses Hg instead of Py.

revlog: export symbol of indexType The idea is to wrap the index object with rust-cpython. I haven't tried it, but it should be doable. We'll probably need a better interface than raw function pointers to do more in Rust.

setup: fix path of rust source dependencies Appears that they were screwed up while moving the source files around. Since rust-direct-ffi sources are automatically added to the depends list, we only need to list hg-core sources.

rebase: remove now-unnecessary arguments to _abort() Since _abort() is now an instance method, we don't need to pass state that's available on "self". Differential Revision: https://phab.mercurial-scm.org/D5393

rebase: move abort() onto rebaseruntime The function depends on a lot of the state of rebaseruntime, so it makes sense for it to be an instance method. This will let us remove many of the arguments to the method. That will be done in a later patch. Differential Revision: https://phab.mercurial-scm.org/D5392