revbranchcache: use context manager in _writerevs() to write to file The other _writenames() is a bit complicated to use context manager. Differential Revision: https://phab.mercurial-scm.org/D6292

revbranchcache: factor logic to write names and revs in separate functions Before this patch, the write function was so populated with upto 4 level of indentation, it was hard to understand what's going on. Differential Revision: https://phab.mercurial-scm.org/D6291

tests: make log style a little easier to read in test-copytrace-heuristics.t Revision numbers are much shorter and easier to read (especially compared to the full nodeids that were used here), so I switched to that. That's also what almost all the commands used (e.g. `hg rebase -s . -d 1`). I updated the two instances that used nodeids. I also made some other little cleanups to the log templates. Differential Revision: https://phab.mercurial-scm.org/D6279

tests: avoid cryptic nodeids in tests/test-rename-merge1.t These two nodeids had not been part of any output before, so one can't know which revision they refer to without adding something like `hg log` before them. It turned out that '.^' was equivalent for both of them, so that's what I replaced them with. Differential Revision: https://phab.mercurial-scm.org/D6280

tests: defines aliases for `hg log` calls in test-copytrace-heuristics.t This also makes the test cases more consistent since a few had missed the ":" in "changeset:" that the others used. Differential Revision: https://phab.mercurial-scm.org/D6278

rust-discovery: implementing and exposing stats() This time, it's simple enough that we can do it in all layers in one shot. Differential Revision: https://phab.mercurial-scm.org/D6233

rust-discovery: cpython bindings for the core logic As previously done with the ancestors submodule, testing for the bindings is provided from Python on a trivial case. Differential Revision: https://phab.mercurial-scm.org/D6232

rust-discovery: starting core implementation Once exposed to the Python side, this core object will avoid costly roundtrips with potentially big sets of revisions. This changeset implements the core logic of the object only, i.e., manipulation of the missing, common and undefined set-like revision attributes. Differential Revision: https://phab.mercurial-scm.org/D6231

rust-dagops: roots Unsuprisingly, the algorithm is much easier than for heads, provided we work on a set in the first place. To improve the signature, a trait for set-likes object would be useful, but that's not an immediate concern. Differential Revision: https://phab.mercurial-scm.org/D6230

rust-dagops: range of revisions This is a Rust implementation for what reachableroots2() does if includepath is True. The algorithmic details and performance notes are included in the documentation comment. Our main use case for now is a Rust counterpart of the partialdiscovery object, so we don't really need bindings yet. Differential Revision: https://phab.mercurial-scm.org/D6229

narrow: also warn when not deleting untracked or ignored files Differential Revision: https://phab.mercurial-scm.org/D6265

setdiscovery: fix a few typos Differential Revision: https://phab.mercurial-scm.org/D6263

copies: delete debug message about "unmatched files new in both" Same reasoning as previous patch. Differential Revision: https://phab.mercurial-scm.org/D6251

copies: delete debug message about changes since common ancestor Same reasoning as previous patch. Differential Revision: https://phab.mercurial-scm.org/D6250

copies: delete debug message about search limit I'm about to rewrite mergecopies() and this message will no longer be emitted then. Let's remove the message now to remove a distraction from that patch. Differential Revision: https://phab.mercurial-scm.org/D6249

copies: move early return for "no copies" case a little earlier We can return before the block that prints debug messages. That block will not be run anyway when there are no copies. Differential Revision: https://phab.mercurial-scm.org/D6248

copies: fix up "fullcopy" with missing entries from "diverge" Similar to the previous patch, but this doesn't even affect tests. It does affect tests if you change them to turn on debug logging. I'm fixing it here so reviewers of the later rewrite patch can hard-code debug logging to be on and more easily compare test results. Differential Revision: https://phab.mercurial-scm.org/D6247

copies: fix up "fullcopy" with missing entries from "copy" This is just a workaround similar to the previous one. It will make it easier to follow later patches. Differential Revision: https://phab.mercurial-scm.org/D6246

merge: remove workaround for issue5020 As I explained in the previous commit, I think the filtering added there is a better fix for the issue, so the workaround from 41f6af50c0d8 (merge: fix crash on criss cross merge with dir move and delete (issue5020), 2017-01-31) should no longer be needed. Differential Revision: https://phab.mercurial-scm.org/D6245

copies: don't include copies that are not in source in directory move I've been working on a rewrite of mergecopies(). I compared the output of the rewritten version with the current version. I noticed that between FIREFOX_NIGHTLY_59_END and FIREFOX_BETA_60_BASE in the mozilla-unified repo, there were many copies that the current version detected that the rewritten version did not. One example was js/src/gc/Iteration.h -> js/src/gc/PublicIterators.h. Then I realized that js/src/gc/Iteration.h doesn't even exist in FIREFOX_NIGHTLY_59_END. This patch adds a filtering step for the "fullcopy" dict. It turns out that that change also affects the test for issue5020 in test-merge-criss-cross.t. The 'dm' action no longer happens there. At first I thought that the test case change meant that this patch was broken, but I think it's actually correct tha the 'dm' action should not happen there. The result of the bid merge is still the same. I suspect this filtering is a better solution for the issue than 41f6af50c0d8 (merge: fix crash on criss cross merge with dir move and delete (issue5020), 2017-01-31). I also suspect that it was broken just a few months earlier by a005c33d0bd7 (mergecopies: add logic to process incomplete data, 2016-10-04). Note that bid merge had been enabled for a few years at that point, since 19903277f035 (merge: use bid merge by default (BC), 2014-10-01). This patch is still just a workaround. It will be cleaned up soon (with the rewrite of mergecopies()). But doing this in a separate patch makes later patches easier to understand and gives a place to explain why this is changing. Differential Revision: https://phab.mercurial-scm.org/D6244

tests: add test for issue5343 (grafting with copies) It seems that issue5353 resulted in a lot of tests in test-graft.t, but the bug actually reported in that issue didn't get a test case. This patch adds one for the "move" and one for the "copy" version of it. I also added a "copy+modify" case, to show what should be a merge conflict. I didn't add one for the "backwards" version of it since the comment says that that was already covered by previous work. The tests added by this patch show the broken behavior (the bug is still open). I suspect the results returned from mergecopies() are not expressive enough to fix this issue: it has a dict for copies to merge with, but that can only give one more filename, but here we need two (one for the path on the remote side and one for the path in the merge base). I want to have it tested anyway since I'm about to refactor mergecopies(). Differential Revision: https://phab.mercurial-scm.org/D6242

chistedit: use context manager to set verbose ui I'm still not exactly sure why this is necessary -- perhaps setting it unconditionally would leak this setting in chg invocations. Regardless, this would have looked very out of place as compared to how this setting is done everywhere else, so at least for the sake of style, let's be consistent with the rest of the codebase.

setdiscovery: stop limiting the number of local head we initially send In our testing this limitation provides now real gain and instead triggers pathological discovery timing for some repository with many heads. See inline documentation for details. Some timing below: Mozilla try repository, (~1M revs, ~35K heads), discovery between 2 clones with 100 head missing on each side before: ! wall 1.492111 comb 1.490000 user 1.450000 sys 0.040000 (best of 20) ! wall 1.813992 comb 1.820000 user 1.700000 sys 0.120000 (max of 20) ! wall 1.574326 comb 1.573500 user 1.522000 sys 0.051500 (avg of 20) ! wall 1.572583 comb 1.570000 user 1.520000 sys 0.050000 (median of 20) after: ! wall 1.147834 comb 1.150000 user 1.090000 sys 0.060000 (best of 20) ! wall 1.449144 comb 1.450000 user 1.330000 sys 0.120000 (max of 20) ! wall 1.204618 comb 1.202500 user 1.146500 sys 0.056000 (avg of 20) ! wall 1.194407 comb 1.190000 user 1.140000 sys 0.050000 (median of 20) pypy (~100 heads, 317 heads) discovery between clones with only 42 common heads before: ! wall 0.031653 comb 0.030000 user 0.030000 sys 0.000000 (best of 25) ! wall 0.055719 comb 0.050000 user 0.040000 sys 0.010000 (max of 25) ! wall 0.038939 comb 0.039600 user 0.038400 sys 0.001200 (avg of 25) ! wall 0.038660 comb 0.050000 user 0.040000 sys 0.010000 (median of 25) after: ! wall 0.018754 comb 0.020000 user 0.020000 sys 0.000000 (best of 49) ! wall 0.034505 comb 0.040000 user 0.030000 sys 0.010000 (max of 49) ! wall 0.019631 comb 0.019796 user 0.018367 sys 0.001429 (avg of 49) ! wall 0.019132 comb 0.020000 user 0.020000 sys 0.000000 (median of 49) Private repository (~1M revs, ~3K heads), discovery from a strip subset, about 100 changesets to be pulled. before: ! wall 1.837729 comb 1.840000 user 1.790000 sys 0.050000 (best of 20) ! wall 2.203468 comb 2.200000 user 2.100000 sys 0.100000 (max of 20) ! wall 2.049355 comb 2.048500 user 2.002500 sys 0.046000 (avg of 20) ! wall 2.035315 comb 2.040000 user 2.000000 sys 0.040000 (median of 20) after: ! wall 0.136598 comb 0.130000 user 0.110000 sys 0.020000 (best of 20) ! wall 0.330519 comb 0.330000 user 0.260000 sys 0.070000 (max of 20) ! wall 0.157254 comb 0.155500 user 0.123000 sys 0.032500 (avg of 20) ! wall 0.149870 comb 0.140000 user 0.110000 sys 0.030000 (median of 20) Same private repo, discovery between two clone with 500 different heads on each side: before: ! wall 2.372919 comb 2.370000 user 2.320000 sys 0.050000 (best of 20) ! wall 2.622422 comb 2.610000 user 2.510000 sys 0.100000 (max of 20) ! wall 2.450135 comb 2.450000 user 2.402000 sys 0.048000 (avg of 20) ! wall 2.443896 comb 2.450000 user 2.410000 sys 0.040000 (median of 20) after: ! wall 0.625497 comb 0.620000 user 0.570000 sys 0.050000 (best of 20) ! wall 0.834723 comb 0.820000 user 0.730000 sys 0.090000 (max of 20) ! wall 0.675725 comb 0.675500 user 0.628000 sys 0.047500 (avg of 20) ! wall 0.671614 comb 0.680000 user 0.640000 sys 0.040000 (median of 20)

peer: introduce a limitedarguments attributes When set to True, it signal that the peer cannot receive too larges arguments and that algorithm must adapt. This should only be True for http peer that does not support argument passed as "post". This will be useful to unlock better discovery performance in the next changesets. I am using a dedicated argument because this is not really a usual "capabilities" things. An alternative approach would be to adds a "large-arguments" to all peer, but the http peers. That seemed a bit too hacky to me.

verify: also check full manifest validity during verify runs Before this changes, `hg verify` only checked if a manifest revision existed and referenced the proper files. However it never checked the manifest revision content itself. Mercurial is expecting manifest entries to be sorted and will crash otherwise. Since `hg verify` did not attempted a full restoration of manifest entry, it could ignore this kind of corruption. This new check significantly increases the cost of a `hg verify` run. This especially affects large repository not using `sparse-revlog`. For now, this is hidden behind the `--full` experimental flag.

verify: introduce an experimental --full flag The flag currently has no effect, see next changeset for details. We introduce the flag as experimental to keep the freedom of changing our mind on the final UI. Note: this patch highlight a small but in `hg help`. An option section is generated even if no option are visible.

verify: introduce a notion of "level" Some checks are slower than others, to help the user to run the checks he needs, we are about to introduce new flag to select faster vs deeper runs. This put the scaffolding in place to do this.

tests: split out separate test for issue5020 The test was added to the existing setup in 41f6af50c0d8 (merge: fix crash on criss cross merge with dir move and delete (issue5020), 2017-01-31). I'm about to make a change that affects that test and it's much easier to follow if the test case for issue5020 is a separate test case. The separate test case is based on what mpm provided in comment 12 on the issue. `hg diff -r 41f6af50c0d8^ tests/test-merge-criss-cross.t` after this patch is pretty small (besides the added test). It's probably easier for reviewers to look at that than to try to understand the diff itself (I don't understand it). Differential Revision: https://phab.mercurial-scm.org/D6243

tests: avoid a rename/delete conflict when updating in test-narrow-update.t After the upcoming rewrite of mergecopies(), this test would otherwise (accurately) start warning about "inside/f1 was deleted and renamed". Differential Revision: https://phab.mercurial-scm.org/D6254

tests: delete unused function in test-rename-merge2.t Differential Revision: https://phab.mercurial-scm.org/D6253

tests: make merge conflicts explicit in `hg annotate` tests We were using `true` as merge tool. I think it makes the test easier to understand if we make the conflicts explcit. It also papered over a conflict that shouldn't have been a conflict (just a bug in copy tracing). I've marked that "BROKEN". Differential Revision: https://phab.mercurial-scm.org/D6252

narrow: make warning about possibly dirty files respect ui.relative-paths Differential Revision: https://phab.mercurial-scm.org/D6264

Added signature for changeset 97ada9b8d51b

Added tag 5.0.2 for changeset 97ada9b8d51b

posix: always seek to EOF when opening a file in append mode Python 3 already does this, so skip it there. Consider the program: #include <stdio.h> int main() { FILE *f = fopen("narf", "w"); fprintf(f, "narf\n"); fclose(f); f = fopen("narf", "a"); printf("%ld\n", ftell(f)); fprintf(f, "troz\n"); printf("%ld\n", ftell(f)); return 0; } on macOS, FreeBSD, and Linux with glibc, this program prints 5 10 but on musl libc (Alpine Linux and probably others) this prints 0 10 By my reading of https://pubs.opengroup.org/onlinepubs/009695399/functions/fopen.html this is technically correct, specifically: > Opening a file with append mode (a as the first character in the > mode argument) shall cause all subsequent writes to the file to be > forced to the then current end-of-file, regardless of intervening > calls to fseek(). in other words, the file position doesn't really matter in append-mode files, and we can't depend on it being at all meaningful unless we perform a seek() before tell() after open(..., 'a'). Experimentally after a .write() we can do a .tell() and it'll always be reasonable, but I'm unclear from reading the specification if that's a smart thing to rely on. This matches what we do on Windows and what Python 3 does for free, so let's just be consistent. Thanks to Yuya for the idea.

move: --force flag forcibly moves, not copies

copy: correct synopsis by making SOURCE a required argument

debugrevlog: fix average size computation for empty data (issue6167) If the file has no full snapshot (eg: was always empty), `hg debugrevlog` would fails when trying to compute their average size.

bookmarks: backout the attempt to fix the delete race This backs out 044045dce23a because it broke a bunch of tests on Windows. Yuya's theory is that we still rely on in-memory changelog data to be flushed out of the transaction.

help: add a missing blank line to unhide `revlog-compression` The help was output, but it was elided with "Enabled by default" from the previous item.

bookmarks: actual fix for race condition deleting bookmark This is a simple but efficient fix to prevent the issue tested in `test-bookmarks-corner-case.t`. It might be worth pursuing a more generic approach where filecache learn to depend on each other, but that would not be suitable for stable. The issue is complicated enough that I documented the race and its current solution as inline comment. See this comment for details on the fix.

localrepo: introduce a `_refreshchangelog` method See next changeset for usage and documentation for details.

bookmarks: actually trigger the race deleting bookmark in the test The previous committed version of the test did not triggered the race, but this was hidden by a strange behavior from the test runner. So we are moving the test to a slightly more complex that actually trigger the issue.

test: add some assert in the bookrace extension This cannot hurt to have a bit more security in the test extension.

test: factor out the "wait" logic in bookrace The test is currently not testing the race it is supposed to test. The synchronisation is still valid, but needs to run at a different point. We start with extracting the synchronisation logic for clarity.

test: remove dead code in the bookrace extension This code is the remain of a previous version of the code. It is never ran, so we can remove it.

run-tests: stop matching line for missing feature Before this change, the following unified test input would silently pass $ echo foo foo (false !) After this change, the "foo" output is properly detected as unexpected. The output of an handful of test had to be updated from broken conditional (that ended up working by chance).

cborutil: fix streamencode() to handle subtypes Otherwise the template filter 'cbor' could crash because of bytes subclass: ValueError: do not know how to encode <class 'mercurial.encoding.safelocalstr'>

bookmarks: use correct store for "ambiguity check" I still don't quite know what the check does, but I clearly got it wrong in 526750cdd02d (bookmarks: keep bookmarks in .hg/store if new config set, 2019-05-15). Just compare with the strings we use in @repofilecache and @storecache. These bugs were then copied to the stable branch in c2b83c957621 (localrepo: grab mixedrepostorecache class from 526750cdd02d, 2019-05-20) and 2338bdea4474 (bookmark: also make bookmark cache depends of the changelog, 2019-05-20). As a result, test-wireproto-exchangev2.t is flaky on both branches. This patch fixes that. Differential Revision: https://phab.mercurial-scm.org/D6469

Added signature for changeset c3484ddbdb96

Added tag 5.0.1 for changeset c3484ddbdb96

manifest: add some documentation to _lazymanifest python code It was not particularly easy figuring out the design of this class and keeping track of how the pieces work. So might as well write some of it down for the next person.

manifest: avoid corruption by dropping removed files with pure (issue5801) Previously, removed files would simply be marked by overwriting the first byte with NUL and dropping their entry in `self.position`. But no effort was made to ignore them when compacting the dictionary into text form. This allowed them to slip into the manifest revision, since the code seems to be trying to minimize the string operations by copying as large a chunk as possible. As part of this, compact() walks the existing text based on entries in the `positions` list, and consumed everything up to the next position entry. This typically resulted in a ValueError complaining about unsorted manifest entries. Sometimes it seems that files do get dropped in large repos- it seems to correspond to there being a new entry that would take the same slot. A much more trivial problem is that if the only changes were removals, `_compact()` didn't even run because `__delitem__` doesn't add anything to `self.extradata`. Now there's an explicit variable to flag this, both to allow `_compact()` to run, and to avoid searching the manifest in cases where there are no removals. In practice, this behavior was mostly obscured by the check in fastdelta() which takes a different path that explicitly drops removed files if there are fewer than 1000 changes. However, timeless has a repo where after rebasing tens of commits, a totally different path[1] is taken that bypasses the change count check and hits this problem. [1] https://www.mercurial-scm.org/repo/hg/file/2338bdea4474/mercurial/manifest.py#l1511

tests: demonstrate broken manifest generation with the pure module This will be fixed next. But I don't fully understand how 'b.txt' is actually removed properly in the second test, given what's broken. Also, I'm not sure why 'bb.txt' is flagged as not being in the manifest, when it clearly appears to be.

bookmark: also make bookmark cache depends of the changelog Since the changelog is also used during the parsing of bookmark data, it should be listed as a file cache dependency. This fix the race condition we just introduced a test for. This is a simple fix that might lead bookmark data to be invalidated more often than necessary. We could have more complicated code to deal with this race in a more "optimal" way. I feel it would be unsuitable for stable. In addition, the performance impact of this is probably minimal and I don't foresee the more advanced fix to actually be necessary.

localrepo: grab mixedrepostorecache class from 526750cdd02d On default, Martin von Zweigbergk <martinvonz@google.com> introduced a more advance filecache decorator. I need this decorator to fix a bug on stable. So I am grafting the relevant part of 526750cdd02d.

bookmark: add a test for a race condition on push Bookmark pointing to unknown nodes are ignored. Later these ignored bookmarks are dropped when writing the file back on disk. On paper, this behavior should be fine, but with the current implementation, it can lead to unexpected bookmark deletions. In theory, to make sure writer as a consistent view, taking the lock also invalidate bookmark data we already loaded into memory. However this invalidation is incomplete. The data are stored in a `filecache` that preserve them if the bookmark related file are untouched. In practice, the bookmark data in memory also depends of the changelog content, because of the step checking if the bookmarks refers to a node known to the changelog. So if the bookmark data were loaded from an up to date bookmark file but filtered with an outdated changelog file this go undetected. This condition is fairly specific, but can occurs very often in practice. We introduce a test recreating the situation. The test comes in an independant changeset to show it actually reproduce the situation. The fix will come soon after. A large share of the initial investigation of this race condition was made by Valentin Gatien-Baron <valentin.gatienbaron@gmail.com>.

test: properly gate a zstd section This part of the test can't run if zstd is not available. This was caught by --pure test (who don't support zstd).

test: update test for expected test output In 1fac9b931d46 as new test session was introduced. It did not take in account some part that only ran for pure. The test is now fixed.

sslutil: fsencode path returned by certifi (issue6132) By inspection, this is the only codepath that could be returning a string instead of a bytes on Python 3.

commit: allow --interactive to work again when naming a directory (issue6131)

parser: fix crash by parsing "()" in keyword argument position A tree node can be either None or a tuple because x=("group", None) is reduced to x[1].

Added signature for changeset 07e479ef7c96

Added tag 5.0 for changeset 07e479ef7c96

hghave: deal with "rc" release Without this change, 5.0rc0 is not recognised as 5.0

narrow: send specs as bundle2 data instead of param (issue5952) (issue6019) Before this patch, when ACL is involved, narrowspecs are send as bundle2 parameter for narrow:spec bundle2 part. The limitation of bundle2 parts are they cannot send data larger than 255 bytes. Includes and excludes in narrow are not limited by size and they can grow over 255 bytes. This patch introduces a new mandatory bundle2 part and send narrowspecs as data of that. The new bundle2 part is introduced to keep things cleaner and easy to distinguish related to backward compatibility. The part is mandatory because without server's narrowspec, the local ACL narrow repo won't work. This patch makes clients compatible with servers which have older versions. However I left a comment that we should drop the other bundle2 part soon as that's broken and people should not rely on that. I named the new bundle2 part 'Narrow:responsespec' because: 1) Capital 'N' to make it mandatory 2) 'Narrow:spec' cannot be used because bundle2 enforces that there should not be two different parts which resolve to same name when lowercased. 3) reponsespec clears that they are specs which are send as reponse by the server While I was here, I renamed `narrowhgacl` section to `narrowacl` as suggested by idlsoft@ and martinvonz@. Differential Revision: https://phab.mercurial-scm.org/D6310

inno: bump keyring to 18.0.1 to avoid AttributeError (issue6043) The error seems to be harmless, because it happens after closing the connection. For whatever reason, this isn't bundled with the Wix installer. https://github.com/jaraco/keyring/issues/386 https://bitbucket.org/Mekk/mercurial_keyring/issues/63/attributeerror-during-process-finish-with

context: check file exists before getting data from _wrappedctx overlayworkingctx class is used to do in-memory merging. The data() function of that class has logic to look for data() in the wrappedctx if the file data in cache is empty and if the file is dirty. This assumes that if a file is dirty and cache has empty data for it, it will exists in the _wrappedctx. However this assumption can be False in case when we are merging a file which is empty in destination. In these cases, the backup file 'foo.orig' created by our internal merge algorithms will be empty, however it won't be present in _wrappedctx. This case will lead us to error like the one this patch is fixing. Let's only fallback to getting data from wrappedctx if cache has 'None' as data. Differential Revision: https://phab.mercurial-scm.org/D6308

tests: show IMM is broken when merging file empty in destination When we are doing in-memory merging, and we are merging a file which is empty in merge destination, it leads to error 'abort: xxx not found in manifest'. Next patch will fix this error. Differential Revision: https://phab.mercurial-scm.org/D6307

buildrpm: bump bundled Python version to 2.7.16 when building for centos{5,6} When building rpm packages for centos 5 and 6, we bundle a mercurial-specific version of Python 2.7 in /opt/python-hg. This change is analogous to 5e947367606c, and bumps the embedded Python version from 2.7.14 (released in 2017) to 2.7.16 (latest as of today).

setup: tweak error message for Python 3 We now have beta support for Python 3. In my opinion, it isn't yet stable enough to allow `pip install Mercurial` to work with Python 3 out of the box: we don't want people accidentally using Mercurial with Python 3 just yet. But I do think we should be more friendly about informing people of their options. This commit tweaks the error message that users see when running setup.py with Python 3. We instruct them about the current level of Python 3 support, point them at the wiki for more info, and give them instructions on how to bypass the check. As part of this, I also changed which version value is printed, as we were printing a named tuple before.

setup: remove set and dict comprehensions Yuya observed in a recent review that it is worthwhile to keep setup.py parseable with Python 2.6 so a useful error message is seen when attempting to run with Python 2.6. This commit removes a set and dict comprehension so setup.py is parseable with Python 2.6.

branchcache: don't verify all nodes while writing nodes are verified either when they are added or used. In case of commits. we will load the whole branchmap, only verify nodes for the branch on which we are committing and then we write. However before this patch, writing the branchmap was validating all the nodes whereas it should not. This patch fixes that. Differential Revision: https://phab.mercurial-scm.org/D6290

setup: properly package distutils in py2exe virtualenv builds Our in-repo py2exe packaging code uses virtualenvs for managing dependencies. An advantage of this is that packaging is more deterministic and reproducible. Without virtualenvs, we need to install packages in the system Python install. Packages installed by other consumers of the system Python could leak into the Mercurial package. A regression from this change was that py2exe packages contained the virtualenv's hacked distutils modules instead of the original distutils modules. (virtualenv installs a hacked distutils module because distutils uses relative path lookups that fail when running from a virtualenv.) This commit introduces a workaround so py2exe packaging uses the original distutils modules when running from a virtualenv. With this change, `import distutils` no longer fails from py2exe builds produced from a virtualenv. This fixes the regression. Furthermore, we now include all distutils modules. Before, py2exe's module finding would only find modules there were explicitly referenced in code. So, we now package a complete copy of distutils instead of a partial one. This is even better than before. # no-check-commit foo_bar function name

merge: forgot to pull before release These two changes should be part of 5.0, but we are fine leaving them out of the RC.

debugdiscovery: include the number of heads in all sets We already displayed information about heads of the common set that are either local or remote heads. We now also do so for heads of the common set that are both local and remote heads too. This is useful because various step in the set discovery algorithm have head specific optimizations.

recover: add a --[no-]verify flag For trivial cases, the cost of the verify run after `hg recover` is getting in the way. In addition for very large repositories, the cost is simply too high to be paid, making `hg recover` an unusable commands. We introduce a --verify flag, set by default. If is automatically associated with a --no-verify flag that one can use to skip the verify step. We might consider changing the default behavior in the future. However this is out of scope for this series.

Added signature for changeset 4a8d9ed86475

Added tag 5.0rc0 for changeset 4a8d9ed86475

merge: default into stable for release candidate

bundle2: handle compression in _forwardchunks _forwardchunks is used to compensate for getbundle protocol deficits. Since it transparently decodes the payload, it also needs to remove the corresponding compression parameter in case the server decides to send one. This the wire protocol part of issue 5990. Differential Revision: https://phab.mercurial-scm.org/D6182

changelog: parse copy metadata if available in extras This lets read back the copy metadata we just started writing. There are still many places left to teach about getting the copy information from the changeset, but we have enough ({file_copies}, specifically) that we can add it now and have some test coverage of it. Differential Revision: https://phab.mercurial-scm.org/D6186

copies: add config option for writing copy metadata to file and/or changset This introduces a config option that lets you choose to write copy metadata to the changeset extras instead of to filelog. There's also an option to write it to both places. I imagine that may possibly be useful when transitioning an existing repo. The copy metadata is stored as two fields in extras: one for copies since p1 and one for copies since p2. I may need to add more information later in order to make copy tracing faster. Specifically, I'm thinking out recording which files were added or removed so that copies._chaincopies() doesn't have to look at the manifest for that. But that would just be an optimization and that can be added once we know if it's necessary. I have also considered saving space by using replacing the destination file path by an index into the "files" list, but that can also be changed later (but before the feature is ready to release). Differential Revision: https://phab.mercurial-scm.org/D6183

revsetbenchmark: add some simpler revset for heads and roots This revset might leverage compiled code in the future so lets start to track them.

repoview: flag `server.view` as experimental Ideally, the non-experimental version of `experimental.extra-filter-revs` will cover the use case for `server.view=immutable` well enough than having to have this dedicated configuration. Since `server.view` is not part of any release, I would prefer to have it marked as experimental to avoid having it to support it for ever.

repoview: move subsettable in a dedicated module The dictionary got moved in `branchmap` to avoid import cycle. However, we are about to needs it in repoview too. So we introduce a now module to define that that mapping.

upgrade: support upgrade to/from zstd storage (issue6088) Now that we have an official config option for a shiny format improvement, we better make it simple to migrate to/from it.

compression: introduce an official `zstd-revlog` requirement This requirement supersedes `exp-compression-zstd`. However, we keep support for the old requirement. Strictly speaking, we do not need to add a new requirement, there are no logic change making "new" repo incompatible with mercurial client using a version that support `exp-compression-zstd`. The choice to introduce a new requirement is motivated by the following: * The previous requirement was explicitly "experimental". Using it by default could confuse users. * adding support for a hypothetical third compression engine will requires new code, and will comes with its own requirement tool. * We won't use it as the default for a while since I do not think we support zstd on all platform. I can imagine we'll gain another (unrelated but on my default) requirement by the time we turn this zstd by default.

compression: only declare revlog support for available engine Even if we know that an engine support revlog compression, the Mercurial process still won't support it if the compression engine is not available.

branchcache: lazily validate nodes in iteritems() This saves ~0.30 sec on creating a new branch on our internal repo. Differential Revision: https://phab.mercurial-scm.org/D6236

branchcache: only iterate over branches which needs to be verified Otherwise we loop over all the branches and call _verifybranch() even if not required. Differential Revision: https://phab.mercurial-scm.org/D6240

branchcache: fix the docstring of _verifybranch() Initially the function was designed to support verifying all branches but later I decided to have a separate function. I forget to remove the doc related to that. Thanks to Yuya for spotting this in review. Differential Revision: https://phab.mercurial-scm.org/D6239

branchcache: don't verify while creating a copy The copy will not be mark as verified, so there is no need to verify nodes. Thanks to Yuya who spotted this while reviewing. Differential Revision: https://phab.mercurial-scm.org/D6238

pull: improved message issued in case of failed update When running `hg pull --update`, the update step may fail. Nothing in the error message help to understand the abort is related to the secondary step (update) instead of the primary step (pull). We now add some information to the error message to clarify it comes from the update part. It is useful in various situation (uncommitted changes blocking the update, update to hidden destination, etc...) The pull output is updated from: $ hg pull ../repo-Bob --rev 956063ac4557 --update pulling from ../repo-Bob searching for changes adding changesets adding manifests adding file changes added 2 changesets with 0 changes to 2 files (+1 heads) (2 other changesets obsolete on arrival) abort: filtered revision '6'! to: $ hg pull ../repo-Bob --rev 956063ac4557 --update pulling from ../repo-Bob searching for changes adding changesets adding manifests adding file changes added 2 changesets with 0 changes to 2 files (+1 heads) (2 other changesets obsolete on arrival) abort: cannot update to target: filtered revision '6'! (I am not sure why the actual error, "filtered revision '6'", is not using the more modern format mentioning the obsolescence fate of '6')

pull: deal with locally filtered changeset passed into --rev Nowadays, it is possible to explicitly pull a remote revision that end up being hidden locally (eg: obsoleted locally). However before this patch, some internal processing where crashing trying to resolve a filtered revision. Without this patches, the pull output result a confusing output: $ hg pull ../repo-Bob --rev 956063ac4557 pulling from ../repo-Bob searching for changes adding changesets adding manifests adding file changes added 2 changesets with 0 changes to 2 files (+1 heads) (2 other changesets obsolete on arrival) abort: 00changelog.i@956063ac4557828781733b2d5677a351ce856f59: filtered node!

absorb: aborting if another operation is in progress This increases safety of using absorb by both aborting when another operation is in progress (since the absorption could confuse any other command a lot) and holding the locks throughout the reading of the working directory (for which changes to absorb) and the reading of the repo (for which changes to absorb into). Differential Revision: https://phab.mercurial-scm.org/D6237

repoview: keep the branchmap cache for the `served.hidden` view warm For the same reason we want to keep the cache for the `served` view up to date, we want to also keep the `served.hidden` view up to date. If some processes with a readonly access to the repo needs to access it, we better have the cache warm to avoid computing the same data over and over and over. In most case (no secret changesets), the "served" and "served.hidden" set will be identical and no cache will actually have to be updated.

repoview: introduce a filter for serving hidden changesets There are multiple usecase for being able to explicitly view or pull obsolete from a server. We need to be able to do so without exposing the secret changesets. We introduces a dedicated repository "view" to do so. Way to expose this "view" to the user will come later. To keep a behavior consistent with expected client/server behavior, the general idea is for the obsolete access to be explicitly requested by the code generating the request. In addition, the will be server side configuration to restrict the access to this feature.

repoview: fix conditional around unserved changesets The conditional could lead to wrong computation since we have more unserved changesets than just the "secret" phase. In addition, now that we have efficient caching of phased changesets, we don't need the conditional anymore.

branch: return early if we find an open named branch apart from default The current code builds a list of all the open named branches except default and then bool that. This is mostly fine until you get a repo which has thousands of named branches. Differential Revision: https://phab.mercurial-scm.org/D6211

branchcache: don't verify closed nodes in _branchtip() We only do membership testing there. Differential Revision: https://phab.mercurial-scm.org/D6210

branchcache: don't verify closed nodes in iteropen() We expect that the nodes passed to iteropen() will be verified. We are only testing for membership in closed nodes set, so we don't need to verify the whole closed nodes set. This will speed up calculating branchheads() when there are lot of closed nodes related to other branches. Differential Revision: https://phab.mercurial-scm.org/D6209

branchcache: lazily validate nodes from the branchmap On my personal hg-repository with 365 entries in .hg/cache/branch2, following are the numbers for perfbranchmapload. Before this patch: ! wall 0.000866 comb 0.000000 user 0.000000 sys 0.000000 (best of 2680) ! wall 0.001525 comb 0.000000 user 0.000000 sys 0.000000 (max of 2680) ! wall 0.001107 comb 0.001097 user 0.001086 sys 0.000011 (avg of 2680) ! wall 0.001104 comb 0.000000 user 0.000000 sys 0.000000 (median of 2680) With this patch: ! wall 0.000530 comb 0.000000 user 0.000000 sys 0.000000 (best of 4240) ! wall 0.001078 comb 0.000000 user 0.000000 sys 0.000000 (max of 4240) ! wall 0.000696 comb 0.000693 user 0.000677 sys 0.000017 (avg of 4240) ! wall 0.000690 comb 0.000000 user 0.000000 sys 0.000000 (median of 4240) On our internal repository with ~20k entries in branchcache, I see improvement from 0.125 sec to 0.066 sec which is 47% speed up. The above are the numbers of perfbranchmapload which shows how much time we saved by not validating the nodes. But we need to validate some nodes. Following are timings of some mercurial operations which have speed up because of this lazy validation of nodes: No-op `hg update` on our internal repository (Avg on 4 runs): Before: 0.540 secs After: 0.430 secs Setting a branch name which already exists without --force (Avg of 4 runs): Before: 0.510 secs After: 0.250 secs I ran the ASV performance suite and was unable to see any improvements except there was improvement of perfdirstatewrite() on netbeans which I think was not related. I looked into the commit code, the command which I am trying to speedup, it looks like it uses revbranchcache to update the branchcache. Differential Revision: https://phab.mercurial-scm.org/D6208

branchcache: add functions to validate changelog nodes This patch adds functions to validate closed nodes, validate nodes for a certain branch and for all the branches. These functions will be used in upcoming patches. Differential Revision: https://phab.mercurial-scm.org/D6207

readline: provide styled prompt to readline (issue6070) Differential Revision: https://phab.mercurial-scm.org/D6168

copies: move comment about implementation of mergecopies() to end When you start reading about mergecopies(), you want to know what it is, not that there are different implementations depending on config, so this patch moves that comment to the end. By the way, we don't seem to define what "copytracing" is. I'm just leaving it that way because I don't know what it is myself. It seems to be referred to only on mergecopies() (and not in pathcopies(), for example), so maybe "copytracing" is supposed to be exactly what mergecopies() does? Differential Revision: https://phab.mercurial-scm.org/D6235

remotefilelog: return expected type from copies overrides copies._computeforwardmissing() and copies._computenonoverlap() return sets, so the overrides should also do that. Differential Revision: https://phab.mercurial-scm.org/D6234

changelog: extract a _string_unescape() to mirror _string_escape() We use our own _string_escape() to encode the "extras" field. Then we use codecs.escape_decode() to escape it. But there's also a little workaround for dealing with escaped text that looks like octal numbers since the fix for https://bz.mercurial-scm.org/show_bug.cgi?id=3156. This patch extracts the call to codecs.escape_decode() along with the fix for octal numbers and puts it in a _string_unescape(). It also updates the test to check for the octal-number case from the aforementioned bug. As you may have suspected, I want to be able to reuse this new function later. Differential Revision: https://phab.mercurial-scm.org/D6184

copies: extract function for deciding whether to use changeset-centric algos We'll eventually have a "experimental.copies.read-from=changeset-only" option too and I don't want to spread the logic for determining if we should use changeset-centric of filelog-centric algorithms. Differential Revision: https://phab.mercurial-scm.org/D6163

getrenamedfn: get copy data from context object if configured The function returned from getrenamedfn() calls filelog.renamed(). That won't work when storing copy metadata in the changeset. I've just switched to a simple implementation here. We may or may not need to optimize it later, possibly by optimizing the callers. No more tests fail with "--extra-config-opt experimental.copies.read-from=compatibility)" than they did before this patch. Differential Revision: https://phab.mercurial-scm.org/D6162

branchmap: implement __contains__() We have good occurences of `if branch in branchmap()` in our code. If __contains__() is not implemented then it will use __iter__() to find whether the element exists or not which is not good. I am bit confused that whether I should move existing callers to hasbranch() or this patch is a good way. Differential Revision: https://phab.mercurial-scm.org/D6206

branchmap: prevent using __getitem__() in branchheads() branchheads() can directly use self._entries instead. Differential Revision: https://phab.mercurial-scm.org/D6205

branchmap: dynamically resolve type of branchcache class This is required to support subclassing. Thanks to Yuya for suggesting this in D6151. Differential Revision: https://phab.mercurial-scm.org/D6204

copies: print list of divergent renames in sorted order The current order is not obvious to the user (and I may change it in a later patch). Lexicographical order seems like the obvious choice here. Differential Revision: https://phab.mercurial-scm.org/D6227

import: suppress accept: header irccloud and others may perform content negotation and reject (406) if the accept header does not match the mime type of the object

rebase: fix bug that prevented dry-run rebases from printing failures As far as I can tell it should be fine to unconditionally skip _prepareabortorcontinue if we're in the process of raising an Abort here. Differential Revision: https://phab.mercurial-scm.org/D6226

rebase: demonstrate bug in dry-run mode which causes cycles to not be reported Differential Revision: https://phab.mercurial-scm.org/D6225

test: minor cleanup to test-server-view.t While looking into adding error output in this test, I did some cleanup.

repoview: improve documentation for `repo.filtered` method I am sitting next to Joerg Sonnenberger and we are discussion his experience with repoview. This first effect of this discussion is this documentation clarification.

revset: short docstring for checkstatus This is where all the action happens for the status-related revsets, and a little documentation doesn't hurt.

discovery: stop direct use of attribute of partialdiscovery Instead of accessing `undecided` directly for ui display purposes, we introduce a `stats()` method that could be extended in the future with more interesting information. This is in preparation for a forthcoming Rust version of this object. Indeed, attributes and furthermore properties are a bit complicated for classes in native code. We could go further and rename `undecided` to mark it private, but `_undecided` is already taken as support for `_undecided` lazyness.

overlayworkingctx: remove misleading trailing slash from directory pattern The paths passed into the matcher are normalized (this applies to include patterns and regular patterns, and to both glob kind and path kind), so the regex for input "foo/" ended up being "foo(?:/|$)". Once we have a (recursive) pattern kind only for directories, we could switch to that here and remove the "mfiles[0] == path" check. Until then, let's at least make it not misleading. Differential Revision: https://phab.mercurial-scm.org/D6224

overlayworkingctx: fix file/dir audit to be repo-relative Before this patch, test-rebase-inmemory.t would stop erroring out about the conflict if you added a "cd a" before line 252. That was because a glob matcher (which are relative) was unintentionally used. That happened because the matcher was given "include" patterns (not regular patterns), and "include" patterns are always glob by default (i.e. unless you write them including the kind prefix). IOW, the "default='path'" argument passed to ctx.match() was ignored. Differential Revision: https://phab.mercurial-scm.org/D6223

messages: replace some instances of "folder" by "directory" I'm pretty sure this is our preferred term. Differential Revision: https://phab.mercurial-scm.org/D6222

match: fix re2 compability broken in 2e2699af5649 When using re2, we call test_match() instead of match() on the compiled regex object. While match() returns a matcher object or None, test_match() returns True or False. So since 2e2699af5649 running test on a machine with a re2 install fails in many places. Instead we make the code a bit more general and everything goes back to normal.

py3: add b'' prefixes to new doctests in match.py # skip-blame as just b'' prefixes Differential Revision: https://phab.mercurial-scm.org/D6221

py3: add one new passing test found by buildbot Differential Revision: https://phab.mercurial-scm.org/D6220

cext: cast s# arguments of Py_BuildValue() to Py_ssize_t The doc doesn't state that "s#" of Py_BuildValue() is controlled by PY_SSIZE_T_CLEAN (unlike the one for PyArg_ParseTuple()), but actually it's switched to Py_ssize_t. https://docs.python.org/2/c-api/arg.html#c.Py_BuildValue https://github.com/python/cpython/blob/2.7/Python/modsupport.c#L432 Follow up for b01bbb8ff1f2 and 896b19d12c08.