repoview: don't crash if mergestate points to non-existent node Differential Revision: https://phab.mercurial-scm.org/D9078

tests: demonstrate crash caused by pinning of non-existent mergestate node If the mergestate somehow contains a reference to a non-existent node, it causes most commands to crash since aaeccdb6e654 (repoview: pin revisions for `local` and `other` when a merge is active, 2020-09-04). This patch adds a test showing that. Differential Revision: https://phab.mercurial-scm.org/D9077

git: also convert timezone to int (issue6359) Credit to moshez for testing this in the wild. Differential Revision: https://phab.mercurial-scm.org/D9062

chg: fallback to original hg if stdio fds are missing If stdio fds are missing (ex. fd 0 is not present), chg might open fds that take the numbers 0, and attachio would send the wrong fds to the client, which might cause unwanted behaviors. Avoid that by detecting the missing fds and falling back to the original hg. Differential Revision: https://phab.mercurial-scm.org/D9058

locking: remove support for inheriting locks in subprocess This seems to have been added for merge driver, and since merge driver is now gone... Differential Revision: https://phab.mercurial-scm.org/D9053

mergedriver: delete it The merge driver code was added in late 2015. I think '406a654b::6f045b56 & user("sid0")' is a reasonable revset for finding the relevant commits, including preparation for it. The code is very poorly tested, which makes it very hard to maintain. It seems it's only used by FB and they don't use this code base anymore, so let's remove the code to make it easier for us to maintain our product. Differential Revision: https://phab.mercurial-scm.org/D9041

rust-cpython: switch logging facade from `simple_logger` to `env_logger` `simple_logger` is just too simple. `env_logger` supports logging to `stderr`, and logging filtering, for example, which are becoming necessary now. The project is nicely active. Differential Revision: https://phab.mercurial-scm.org/D8990

git: pass `id` attribute of `pygit2.Tree` object `pygit2`'s Repository object expects an instance of `Oid` to return the corresponding object, instead of the object itself. After this change and D9062, `hg commit -i` seems to work, unless it hits a case folding assertion (ie trying to add a file like `README.md`). Differential Revision: https://phab.mercurial-scm.org/D9068

branchmap: add a cache validation cache, avoid expensive re-hash on every use In a pathological `hg log` case, we end up executing the branchmap validity checking twice per commit displayed. Or maybe we always do, and I just noticed because it's really slow in this repo for some reason. Before: ``` Time (mean ± σ): 9.816 s ± 0.071 s [User: 9.435 s, System: 0.392 s] Range (min … max): 9.709 s … 9.920 s ``` After: ``` Time (mean ± σ): 8.671 s ± 0.078 s [User: 8.309 s, System: 0.392 s] Range (min … max): 8.594 s … 8.816 s ``` Differential Revision: https://phab.mercurial-scm.org/D9023

merge with stable

run-test: allow relative path in `--blacklist` and `--whitelist` (issue6351) When specifying a test with `--blacklist` or `--whitelist` with path relatives to the repository root (eg: `tests/test-check-commit.t`) the file is not taken into account. It only works when the name of the test is given. It would be better if `--blacklist` and `--whitelist` behaviors where compatible with `--test-list`. This patch allows to use relative path with `--blacklist` and `--whitelist` while staying compatible with the old behavior by checking the test relative path in addition to its name. Differential Revision: https://phab.mercurial-scm.org/D9024

rebase: stop clearing on-disk mergestate when running in memory In-memory merge no longer uses on-disk mergestate since 19590b126764 (merge: use in-memory mergestate when using in-memory context, 2020-09-15). This patch removes the clearing from two places in the in-memory rebase code. I've verified that the one with a TODO was indeed fixed by the aforementioned commit. The other instance doesn't seem to have any test coverage, but I'm pretty confident it's also safe to remove. Differential Revision: https://phab.mercurial-scm.org/D9057

rebase: delete unused p1 argument to _concludenode() Unused since a0192a03216d (rebase: remove now unnecessary logic to allow empty commit when branch changes, 2020-07-09). Differential Revision: https://phab.mercurial-scm.org/D9055

rebase: fix an inconsistent hyphenation in a debug message We used "rebasing on disk" but "rebasing in-memory". I believe the former is correct, so I used that. Differential Revision: https://phab.mercurial-scm.org/D9054

strip: with --keep, consider all revs "removed" from the wcp (issue6270) The current code was blink to change from other branches when stripping merges that are ancestors of the working copy parents.

test-strip: display more information highlight buggy behavior When using `hg strip --keep` on a set of changeset that contains a merge, the set of file considered when rebuilding the dirstate is missing files updated by changeset that are not directly inside the `new-parent::old-parent` range. We start with updating the test with new output highlighting the issue. This issue was spotted because that dirstate inconsistency made the test flaky. The new command make the test less flaky (but still wrong).

changing-files: implement equality checking This makes debugging simpler.

changing-files: document the various sets The content of some sets (eg: added, removed) is not always obvious so we document them.

merge with stable

share: do not write working-copy requirements twice Follows up d252f51ab032. Maybe it is a copy-paste error.

py3: fix formatting of LookupError for workingctx Spotted while writing broken tests for "hg grep -fr'wdir()'". basectx._fileinfo() raises ManifestLookupError(self._node, ..), but _node of the workingctx is None for historical reasons.

phabricator: fix loadhgrc() override broken by D8656 This fixes some failing `test-check*` tests (at least `test-check-code.t` was broken). Differential Revision: https://phab.mercurial-scm.org/D9044

mergestate: move commit() from base class to on-disk subclass The commit message doesn't do anything in the in-memory case, and it's `_makerecords()` is also not needed in that case, so let's move it down to the on-disk subclass (i.e. the `mergestate` class). Sorry I missed this earlier. Differential Revision: https://phab.mercurial-scm.org/D9043

mergestate: make in-memory mergestate not clear on-disk mergestate on reset() Oops, I thought I had remove the `rmtree()` call earlier. Maybe I accidentally got it back in a histedit or something. Differential Revision: https://phab.mercurial-scm.org/D9042

merge: use in-memory mergestate when using in-memory context This is my version of Augie's D8568. It makes it so we don't touch the mergestate on disk when using an in-memory context. The reason that I want this is not the same as the reason that Augie write his patch (though I agree with that reason too). My hope is to make in-memory rebase not fall back to on-disk rebase when there are conflict. I plan to do that by adding a `overlayworkingctx.reflect_in_workingcopy()`. The idea is that that will update the working copy, the dirstate and the mergestate as necessary. Differential Revision: https://phab.mercurial-scm.org/D9040

mergestate: extract a base class to be shared by future memmergestate This extracts a new base class from `mergestate` and leaves all the vfs-touching code in `mergestate`. Differential Revision: https://phab.mercurial-scm.org/D9039

mergestate: extract overridable methods for making/restoring file backups In-memory merge currently still uses on-disk mergestate. That's mostly harmless (I don't think I've seen any problem it's actually caused). It's still a little weird. I'm planning to add an in-memory mergestate, which will be used with `overlayworkingctx`. This patch prepares for that by extracting that logic, so it's easier to change per subclass. Differential Revision: https://phab.mercurial-scm.org/D9038

mergestate: initialize all properties in __init__() This is hopefully not very controverial. I found the initialization before this patch unorthodox. It wasn't clear which properties the object was supposed to have. Differential Revision: https://phab.mercurial-scm.org/D9037

mergestate: remove unnecessary clearing of `localctx` and `otherctx` As noted in the previous commit, there are no callers that (re-)use the instance after calling `reset()`. There are also no callers that call `_read()` after doing anything with the instance (it's only called right after an instance is created). If reviewers feel that this is too risky, I can extract and reuse the poisoning code that indygreg once added for poisining repo instances. Differential Revision: https://phab.mercurial-scm.org/D9036

mergestate: move most of of reset() into start() `ms.reset()` has somehow become a combination of two different things: 1. A constructor-like function creating an empty instance 2. A call to `shutil.rmtree()` to clear the mergestate. It seems that all callers now care only about the latter (since we changed one caller to use the new `ms.start()` method instead). Let's move the code for the former into `start()`, since that's the only place it's needed. Differential Revision: https://phab.mercurial-scm.org/D9035