upgrade: speed up when we have only nodemap to downgrade Similar to what we do on upgrade, if we have only persistent-nodemap to downgrade we will just delete the nodemap files and update repository requirements instead of processing all the revlogs. After downgrade, we are left with unrequired docket and transaction files which seems fine but can work on deleting them if someone feels we should. Differential Revision: https://phab.mercurial-scm.org/D9992

upgrade: write nodemap for manifests too In 98e39f04d60e I assumed that writing nodemap for manifests was not desirable and stopped writing it during upgrade. However in recent discussion with Pierre-Yves, I learnt that that's not true. Differential Revision: https://phab.mercurial-scm.org/D9991

windows: fix parsing of version number to match format from D9955 Differential Revision: https://phab.mercurial-scm.org/D10061

build: make version from .hg_archival.txt consistent with that from .hg/ D9955 changed the version format to replace "-" by "." and to add "hg" before the number representing the distance from the latest tag. However, it missed the "hg" string and added an extra "." to the version string we produce when there's a `.hg_archival.txt`. This patch makes it consistent. Differential Revision: https://phab.mercurial-scm.org/D10060

helptext: fix a recent typo Differential Revision: https://phab.mercurial-scm.org/D10033

re2: byteify some regex used to get check re2 availability Changeset 687b865b95ad failed to properly byteify this because it was a raw string. This went undetected so far because re2 does not seems to be widely tested. Differential Revision: https://phab.mercurial-scm.org/D10064

packaging: add Provides: python3-mercurial and Homepage to debian package There are other packages that depend on python3-mercurial, like debian's mercurial-git, so we should mark ourselves as providing it. I compared the control file we generate to the one that the debian maintainers generate, and noticed several differences: - the Homepage bit. I included this, because why not - a more robust Suggests list that includes a graphical merge tool - a more robust Breaks list - debian's Recommends openssh-client, we only Recommends ca-certificates - a split into `mercurial` and `mercurial-common` (and possibly others?) - a slightly different description Differential Revision: https://phab.mercurial-scm.org/D9983

bundle2: pass the operation source down to the changegroup This is currently not used by anything in core (and redundant with the url), the real source information is much more useful. This is going to be used in sidedata exchange patches coming soon. Differential Revision: https://phab.mercurial-scm.org/D9986

gendoc: use an empty comment so aliases are separated from previous elements For commands like `hg bookmarks`, where there's no `[+] marked option can be specified multiple times`, this causes the final option in the option list to not be the parent of the aliases definition. The aliases section is thus marked as a blockquote like on commands that do have text separating the option list and the aliases definition. Differential Revision: https://phab.mercurial-scm.org/D9990

gendoc: add support for loading extensions from config settings We manage our installation and ship some extensions, enabled by default for our users, that are in hgext3rd or other directories not scanned by this tool by default. We want to generate docs during the build process, and having those docs include the extensions that users don't have to manually enable is desirable. This is *not* desirable for the normal build process, however, and should never be enabled by default. Differential Revision: https://phab.mercurial-scm.org/D9989

gendoc: support defaults on customopts a bit better Without this, a customopt will very likely render like this: ``` -foo does foo (default: <hgext.myextension.MyOpt object at 0x7f31...>) ``` I copied this logic from how this is handled in mercurial/help.py. Differential Revision: https://phab.mercurial-scm.org/D9988

softstrip: fix a reference to an undefined variable `backupfile` wasn't defined if no backup was requested. Let's set it to `None` by default, which matches what regular `repair.strip()` does. Differential Revision: https://phab.mercurial-scm.org/D9985

softstrip: move _bookmarkmovements() call to where it's needed The call to `_bookmarkmovements()` is unrelated to the backup, so let's move it after. Differential Revision: https://phab.mercurial-scm.org/D9984

tests: fix differing output between py2 and py3 db9e33beb0fb broke the tests because of the difference in bytestring repr between py2 and py3. Rather than backout that change for so little, I figured I'd fix it myself. Hopefully Python 2 supports gets dropped very soon. Differential Revision: https://phab.mercurial-scm.org/D9987

build: fake PEP440 versions If the current version is not exactly a tag, use a local version specifier to fix it up. PEP 440 uses the "+" separator and only allows alphanumeric and dot, so use dot for further separations. Old devel build: 5.7+155-a163cc36d06b New devel build: 5.7+hg155.a163cc36d06b Differential Revision: https://phab.mercurial-scm.org/D9955

tests: add a comment in a test that will hopefully save someone some time I spent at least an hour, probably closer to 1.5, trying to figure out what this was complaining about. Hopefully anyone else in my position will see this note and not waste the time. Differential Revision: https://phab.mercurial-scm.org/D9953

changegroup: don't convert revisions to node for duplicate handling The only consumer can handle revision lists fine. Avoid materializing a range if there are no duplicates as optimization. Differential Revision: https://phab.mercurial-scm.org/D9884

revlog: change addgroup callbacks to take revision numbers Differential Revision: https://phab.mercurial-scm.org/D9883

revlog: change addrevision to return the new revision, not node Differential Revision: https://phab.mercurial-scm.org/D9882

revlog: change addrawrevision to return the revision Differential Revision: https://phab.mercurial-scm.org/D9881

revlog: change _addrevision to return the new revision The node is passed as argument already, so returning it is quite pointless. The revision number on the other is useful as it decouples the caller from the revlog internals. Differential Revision: https://phab.mercurial-scm.org/D9880

rhg: add limited support for the `config` sub-command Only with one argument and no flag. This is mostly for testing. Differential Revision: https://phab.mercurial-scm.org/D9972

rhg: Add support for --config CLI arguments Differential Revision: https://phab.mercurial-scm.org/D9971

rhg: Add support for -R and --repository command-line arguments Differential Revision: https://phab.mercurial-scm.org/D9970

rhg: Replace subcommand boilerplate with a macro This removes some repetition, and will avoid additional repetition in the next commit. Differential Revision: https://phab.mercurial-scm.org/D9969

rhg: Move subcommand CLI arguments definitions to respective modules Differential Revision: https://phab.mercurial-scm.org/D9968

rhg: replace command structs with functions The `Command` trait was not used in any generic context, and the struct where nothing more than holders for values parsed from CLI arguments to be available to a `run` method. Differential Revision: https://phab.mercurial-scm.org/D9967

rust: Use the DisplayBytes trait in config printing This is similar to `std::fmt::Display`, but for arbitrary bytes instead of Unicode. Writing to an abstract output stream helps avoid allocating intermediate `Vec<u8>` buffers. Differential Revision: https://phab.mercurial-scm.org/D9966

log: respect diff.merge in -p output Differential Revision: https://phab.mercurial-scm.org/D9958

diff: extract function for getting possibly re-merged parent to diff against We'll want to reuse the logic that `hg diff --change` with `diff.merge` uses. At least `hg log -p` should reuse it. This patch therefore extracts that code to a function. Differential Revision: https://phab.mercurial-scm.org/D9957

diff: replace --merge option by config option I can't think of any reason you'd want to enable the merge diff on a run-to-run basis; you'd probably either always or never want it set (though I can't see why you'd never want it set). If you have it set, you'll probably also want the same output in `hg log -p` output. Having a single config option for the feature makes sense. Differential Revision: https://phab.mercurial-scm.org/D9956

tagcache: distinguish between invalid and missing entries The TortoiseHg repo has typically not had a newly applied tag accessible by name for recent releases, for unknown reasons. Deleting and rebuilding the tag cache doesn't fix it, though deleting the cache and running `hg log -r $new_tag` does. Eventually the situation does sort itself out for new clones from the server. In an effort to figure out what the issue is, Pierre-Yves David suggested listing these entries in the debug output more specifically. This isn't complete yet- the second test change that says "missing" is more like "invalid", since it was truncated. The problem there is the code that reads the raw array truncates any partial records and then fills it with 0xFF, which signifies that it is missing. As a side note, that means the check for the length when validating an existing entry never fails. Differential Revision: https://phab.mercurial-scm.org/D9811

branching: merge with stable

hooks: add a `auto` value for `hooks.*run-with-plain` That setting restore the behavior pre-5.6. The current HGPLAIN value is simply passed to the hooks. This allow user who needs it to fully mitigate the behavior change introduced in Mercurial 5.7 by restoring the older behavior. Differential Revision: https://phab.mercurial-scm.org/D9982

hooks: introduce a `:run-with-plain` option for hooks This option control if HGPLAIN should be set or not for the hooks. This is the first step to give user some control of the HGPLAIN setting for they hooks. Some hooks (eg: consistency checking) deserve to be run with HGPLAIN, some other (eg: user set visual helper) might need to respect the user config and setting. So both usage are valid and we need to restore the ability to run -without- HGPLAIN that got lost in Mercurial 5.7. This does not offer a way to restore the pre-5.7 behavior yet (respect whatever HGPLAIN setting from the shell), this will be dealt with in the next changeset. The option name is a bit verbose because implementing this highlighs the need for another option: `:run-if-plain`. That would make it possible for some hooks to be easily disabled if HG PLAIN is set. However such option would be a new feature, not something introduced to mitigate a behavior change introduced in 5.7, so the `:run-if-plain` option belong to the default branch and is not part of this series. Differential Revision: https://phab.mercurial-scm.org/D9981

hooks: add some test about HGPLAIN setting and hooks In Mercurial 5.7, hooks are now ran with HGPLAIN set, which is a behavior change in. I could not find explicit test about it so I am adding one. The next changesets will introduce more change to help user mitigate the behavior change when needed. Differential Revision: https://phab.mercurial-scm.org/D9979

hooks: forbid ':' in hook name The `:` character is a special separator in the config and it seems same do to the same for hooks. This is necessary to improve the experience around the HGPLAIN behavior change in 5.7. See next changesets for details. Differential Revision: https://phab.mercurial-scm.org/D9978

rust-status: honor matcher when using the dirstate-only fast-path (issue6483) Differential Revision: https://phab.mercurial-scm.org/D9977

status: add test that shows that the Rust implementation has a bug Differential Revision: https://phab.mercurial-scm.org/D9976

tweak-default: no longer enable the experimental return code The change is quite new and undocumented (since it is experimental) so it seems premature to make it available in tweak default. In addition, I am not sure the new return code are frozen yet (eg: some of the initial feedback have not been incorporated). Before the release I doubled check (probably with Martin) that they were not enabled by default and got replied that they were only enabled in the tests. Have I been aware that they have been also enabled in tweak default I would I have lobbied to delay that. I discovered they were in tweak default from users feedback. They found it an unpleasant and unexpected surprise of 5.7. So I suggest we no enable enable this experimental feature for Mercurial 5.7 and revisit this later, when the feature will be more mature. Differential Revision: https://phab.mercurial-scm.org/D9975

rhg: Parse per-repository configuration Differential Revision: https://phab.mercurial-scm.org/D9964

rhg: Abort based on config on share-safe mismatch Differential Revision: https://phab.mercurial-scm.org/D9963

rhg: Parse system and user configuration at program start … and pass it around up to `Repo::find` Differential Revision: https://phab.mercurial-scm.org/D9962

rust: Parse system and user configuration CLI `--config` argument parsing is still missing, as is per-repo config Differential Revision: https://phab.mercurial-scm.org/D9961

rust: Remove unnecessary check for absolute path before joining `Path::join` does the right thing if its argument is absolute. Differential Revision: https://phab.mercurial-scm.org/D9960

rust: replace read_whole_file with std::fs::read It does the same thing Differential Revision: https://phab.mercurial-scm.org/D9959

wireprotopeer: clarify some variable names now that we allow snake_case "encargsorres" is hard to parse ("encarg sorres" sounds like it might be Spanish to me, and indeed Google Translate tells me that it's Catalan for "order sands"). Let's clarify with some added underscores and longer names. Differential Revision: https://phab.mercurial-scm.org/D9973

copy-tracing: add a --compute flag to debugchangedfiles This will help analysis of possible misbehaving cases. Differential Revision: https://phab.mercurial-scm.org/D9946

bundle2: print "error:abort" message to stderr instead of stdout It seems like the server's message is something you'd like to see even with `--quiet`. It's clearly part of the error. Differential Revision: https://phab.mercurial-scm.org/D9954

narrow: add --no-backup option for narrowing Most of our users at Google use Mercurial on a file system that keeps backups of previous versions of all files, including those in `.hg/`. They therefore don't need a separate backup in the file system when narrowing their repo (which they typically do by running `hg tracked --auto-remove-includes`). Backups can be very slow. `hg strip` already has a `--no-backup` option. This patch adds the same option to `hg tracked --removeinclude/--addexclude`. Differential Revision: https://phab.mercurial-scm.org/D9951

debugdiscovery: add flags to run discovery on subsets of the local repo Generating new repository using strip of local clone is very expensive for large repositories. And such large repository are the most likely to requires debugging around discovery. So we add a simple way to run discovery using provided sets of heads. Differential Revision: https://phab.mercurial-scm.org/D9945

cmdutil: add a missing byte prefix to string introduce in 976b26bdd0d8 The change is missing a the `b'foo'` prefix to make it a bytestring. This lead to a traceback in some third party extension. It is unclear to me why the Mercurial test pass without it. Differential Revision: https://phab.mercurial-scm.org/D9974

relnote: remove the reference to `debugstrip` The `debug` namespace is not intended for end user and advertising it is a path to confusion and trouble. I think we should wait for the `admin` namespace to exists and the command to be available as `admin--strip` before we advertise it. Differential Revision: https://phab.mercurial-scm.org/D9817

branching: merge with stable

upgrade: implement partial upgrade for upgrading persistent-nodemap Upgrading repositories to use persistent nodemap should be fast and easy as it requires only two things: 1) Updating the requirements 2) Writing a persistent-nodemap on disk For both of the steps above, we don't need to edit existing revlogs. This patch makes upgrade only do the above mentioned two steps if we are only upgarding to use persistent-nodemap feature. Since `nodemap.persist_nodemap()` assumes that there exists a nodemap file for the given revlog if we are trying to call it, this patch adds `force` argument to create a file if does not exist which is true in our upgrade case. The test changes demonstrate that we no longer write nodemap files for manifest after upgrade which I think is desirable. Differential Revision: https://phab.mercurial-scm.org/D9936

tests: unquiet a test to show changes in next patch Differential Revision: https://phab.mercurial-scm.org/D9935

nodemap: fix a typo in error message Differential Revision: https://phab.mercurial-scm.org/D9934

revlog: refactor logic to compute nodemap file in separate function I will like to use it one more place. Differential Revision: https://phab.mercurial-scm.org/D9933

nodemap: make `_persist_nodemap` a public function I will like to have a utility function using which we can directly write out nodemap for a repository without going through the recloning process. This function seems like the one containing important pieces for that. Let's make it public. Differential Revision: https://phab.mercurial-scm.org/D9932

engine: 'if not, else' -> 'if, else' I personally feel that ``` if x: pass else: pass ``` is easier to read and edit than ``` if not x: pass else: pass ``` Next patches will add one more if-else clause. Differential Revision: https://phab.mercurial-scm.org/D9931

debugcommands: s/stdin/stdout in debugnodemap help Differential Revision: https://phab.mercurial-scm.org/D9930

errors: use InputError in uncommit extension Differential Revision: https://phab.mercurial-scm.org/D9911

largefiles: properly pass kwargs into url.open The url.open function has acquired a lot of kwargs over the years. When running `hg import http://example.com/hg/diff/1`, since at least a708e1e4d7a8 in March, 2018, the calling sites for url.open try to pass a `sendaccept` parameter that largefiles' override doesn't accept. Currently that stack traces something like this: Traceback (most recent call last): File "/tmp/hgtests.sv744r5t/install/bin/hg", line 59, in <module> dispatch.run() File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/dispatch.py", line 143, in run status = dispatch(req) File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/dispatch.py", line 245, in dispatch status = _rundispatch(req) File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/dispatch.py", line 289, in _rundispatch ret = _runcatch(req) or 0 File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/dispatch.py", line 465, in _runcatch return _callcatch(ui, _runcatchfunc) File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/dispatch.py", line 475, in _callcatch return scmutil.callcatch(ui, func) File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/scmutil.py", line 155, in callcatch return func() File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/dispatch.py", line 455, in _runcatchfunc return _dispatch(req) File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/dispatch.py", line 1259, in _dispatch lui, repo, cmd, fullargs, ui, options, d, cmdpats, cmdoptions File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/dispatch.py", line 913, in runcommand ret = _runcommand(ui, options, cmd, d) File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/dispatch.py", line 1270, in _runcommand return cmdfunc() File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/dispatch.py", line 1256, in <lambda> d = lambda: util.checksignature(func)(ui, *args, **strcmdopt) File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/util.py", line 1867, in check return func(*args, **kwargs) File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/commands.py", line 4184, in import_ patchfile = hg.openpath(ui, patchurl, sendaccept=False) File "/tmp/hgtests.sv744r5t/install/lib/python/mercurial/hg.py", line 181, in openpath return url.open(ui, path, sendaccept=sendaccept) TypeError: openlargefile() got an unexpected keyword argument 'sendaccept' So, just accept and pass along any kwargs of the overridden function.

rhg: add support for share-safe Differential Revision: https://phab.mercurial-scm.org/D9942

rhg: initial support for shared repositories Differential Revision: https://phab.mercurial-scm.org/D9941

rhg: Add basic test with a shared repository Differential Revision: https://phab.mercurial-scm.org/D9940

tests: print the server error log in `test-url-download.t` There was a stray 500 error in CI, but no additional context to know what the issue was. https://foss.heptapod.net/octobus/mercurial-devel/-/jobs/163093 Differential Revision: https://phab.mercurial-scm.org/D9949

procutil: extend gui test to detect wayland session (issue6479)

Added signature for changeset 0e2e7300f430

Added tag 5.7 for changeset 0e2e7300f430

ci-fix: backed out changeset 45afff0f530a This changeset is part of a series that break Continuous integration on python 2 for about a week. As not concrete solution have been found so far the safer seems to back it out until we can figure the details out. Differential Revision: https://phab.mercurial-scm.org/D9948

ci-fix: backed out changeset d4c8b4b90ecb This changeset is part of a series that break Continuous integration on python 2 for about a week. As not concrete solution have been found so far the safer seems to back it out until we can figure the details out. Differential Revision: https://phab.mercurial-scm.org/D9947

relnotes: add entry for `hg diff --merge -c` Differential Revision: https://phab.mercurial-scm.org/D9944

branching: merge with stable

relnotes: copy "next" to "5.7" and clear "next" The same procedure as every year^Wcycle. Differential Revision: https://phab.mercurial-scm.org/D9909

relnotes: elaborate on the side effects of hooks having HGPLAIN=1 Differential Revision: https://phab.mercurial-scm.org/D9943

packaging: replace a documentation reference to `python3` on Windows Sadly, there's not `python3.exe` on Windows. Differential Revision: https://phab.mercurial-scm.org/D9937

diff: suppress `merging foo` output lines when performing a merge diff Differential Revision: https://phab.mercurial-scm.org/D9939

diff: add experimental support for "merge diffs" The way this works is it re-runs the merge and "stores" conflicts, and then diffs against the conflicted result. In a normal merge, you should only see diffs against conflicted regions or in cases where there was a semantic conflict but not a textual one. This makes it easier to detect "evil merges" that contain substantial new work embedded in the merge commit. Differential Revision: https://phab.mercurial-scm.org/D8504

rust: use HgError in ConfigError Differential Revision: https://phab.mercurial-scm.org/D9938

rust: Fold find_root and check_requirements into Repo::find Differential Revision: https://phab.mercurial-scm.org/D9906

rhg: Simplify CommandError based on its use Differential Revision: https://phab.mercurial-scm.org/D9905

rust: replace ToString impls with Display ToString is automatically implementing for everything that implements Display, and Display can avoid allocating intermediate strings. Differential Revision: https://phab.mercurial-scm.org/D9904

rust: use HgError in RevlogError and Vfs Differential Revision: https://phab.mercurial-scm.org/D9897

rust: replace RequirementsError with HgError Differential Revision: https://phab.mercurial-scm.org/D9896

rhg: Print an error message in more cases Differential Revision: https://phab.mercurial-scm.org/D9895

rust: Remove DirstateParseError and ListDirstateTrackedFilesError Use HgError instead. Differential Revision: https://phab.mercurial-scm.org/D9894

rust: Replace DirstatePackError with HgError Differential Revision: https://phab.mercurial-scm.org/D9893

rust: Introduce an `HgError` enum for common error cases Differential Revision: https://phab.mercurial-scm.org/D9892

rust: remove three enums that were identical to `RevlogError` Differential Revision: https://phab.mercurial-scm.org/D9877

rhg: replace `map_*_error` functions with `From` impls Differential Revision: https://phab.mercurial-scm.org/D9876

rust: replace trivial `impl From …` with `#[derive(derive_more::From)]` Crate docs: https://jeltef.github.io/derive_more/derive_more/from.html Differential Revision: https://phab.mercurial-scm.org/D9875

rust: remove `FooError` structs with only `kind: FooErrorKind` enum field Use the enum directly as `FooError` instead. Differential Revision: https://phab.mercurial-scm.org/D9874

rhg: centralize parsing of `--rev` CLI arguments This new module will be the place to implement more of the revset language when we do so. Differential Revision: https://phab.mercurial-scm.org/D9873

rust: Remove hex parsing from the nodemap Separating concerns simplifies error types. Differential Revision: https://phab.mercurial-scm.org/D9864

rust: Make NodePrefix allocation-free and Copy, remove NodePrefixRef The `*Ref` struct only existed to avoid allocating `Vec`s when cloning `NodePrefix`, but we can avoid having `Vec` in the first place by using an inline array instead. This makes `NodePrefix` 21 bytes (with 1 for the length) which is smaller than before as `Vec` alone is 24 bytes. Differential Revision: https://phab.mercurial-scm.org/D9863

churn: count lines that look like diff headers but are not Previously, churn cannot count added lines that start with "++ " or removed lines that start with "-- ". Differential Revision: https://phab.mercurial-scm.org/D9929

rust: Exclude empty node prefixes We presumably don’t want `--rev ""` to select every single revision, even though the empty string is a prefix of all strings. Differential Revision: https://phab.mercurial-scm.org/D9862

rust: Simplify error type for reading hex node IDs If a string is not valid hexadecimal it’s not that useful to track the precise reason. Differential Revision: https://phab.mercurial-scm.org/D9861

rust: replace Node::encode_hex with std::fmt::LowerHex This avoids allocating intermediate strings. Differential Revision: https://phab.mercurial-scm.org/D9860

rhg: Build in release mode on CI This follows e73b40c790ec which made tests use the release executable. With e73b40c790ec but not this, tests are skipped on CI because the executable is missing. Differential Revision: https://phab.mercurial-scm.org/D9907

config: use the right API to access git-submodule Differential Revision: https://phab.mercurial-scm.org/D9923

config: use the right API to access template access Preventing direct access to the underlying dict will help a coming refactoring of `config`. Differential Revision: https://phab.mercurial-scm.org/D9922

config: use the right API to access subrepository section Preventing direct access to the underlying dict will help a coming refactoring of `config`. Differential Revision: https://phab.mercurial-scm.org/D9921

config: test priority involving the command line Differential Revision: https://phab.mercurial-scm.org/D9917

config: test priority involving include Differential Revision: https://phab.mercurial-scm.org/D9916

config: add a test for priority when includes are involved Differential Revision: https://phab.mercurial-scm.org/D9915

errors: use InputError for incorrectly formatted dates Differential Revision: https://phab.mercurial-scm.org/D9913

errors: use StateError more in merge module Differential Revision: https://phab.mercurial-scm.org/D9912

errors: use exit code 40 for when a hook fails A hook can be used for checking inputs, state, configuration, security, etc., so it's unclear which of the existing exit codes to use. Let's instead add one specifically for failed hooks. I picked 40. Differential Revision: https://phab.mercurial-scm.org/D9910

context: add missing manifest invalidation after write in overlayworkingctx This was breaking my merge-diff logic that will be in the next patch. Differential Revision: https://phab.mercurial-scm.org/D9908

wix: tell ComponentSearch that it is finding a directory (not a file) This is to fix an issue we've noticed where fresh installations start at `C:\Program Files\Mercurial`, and then upgrades "walk up" the tree and end up in `C:\Program Files` and finally `C:\` (where they stay). ComponentSearch defaults to finding files, which I think means "it produces a string like `C:\Program Files\Mercurial`", whereas with the type being explicitly a directory, it would return `C:\Program Files\Mercurial\` (note the final trailing backslash). Presumably, a latter step then tries to turn that file name into a proper directory, by removing everything after the last `\`. This could likely also be fixed by actually searching for the component for hg.exe itself. That seemed a lot more complicated, as the GUID for hg.exe isn't known in this file (it's one of the "auto-derived" ones). We could also consider adding a Condition that I think could check the Property and ensure it's either empty or ends in a trailing slash, but that would be an installer runtime check and I'm not convinced it'd actually be useful. This will *not* cause existing installations that are in one of the bad directories to fix themselves. Doing that would require a fair amount more understanding of wix and windows installer than I have, and it *probably* wouldn't be possible to be 100% correct about it either (there's nothing preventing a user from intentionally installing it in C:\, though I don't know why they would do so). If someone wants to tackle fixing existing installations, I think that the first installation is actually the only one that shows up in "Add or Remove Programs", and that its registry keys still exist. You might be able to find something under HKEY_USERS that lists both the "good" and the "bad" InstallDirs. Mine was under `HKEY_USERS\S-1-5-18\Software\Mercurial\InstallDir` (C:\), and `HKEY_USERS\S-1-5-21-..numbers..\Software\Mercurial\InstallDir` (C:\Program Files\Mercurial). If you find exactly two, with one being the default path, and the other being a prefix of it, the user almost certainly hit this bug :D We had originally thought that this bug might be due to unattended installations/upgrades, but I no longer think that's the case. We were able to reproduce the issue by uninstalling all copies of Mercurial I could find, installing one version (it chose the correct location), and then starting the installer for a different version (higher or lower didn't matter). I did not need to deal with an unattended or headless installation/upgrade to trigger the issue, but it's possible that my system was "primed" for this bug to happen because of a previous unattended installation/upgrade. Differential Revision: https://phab.mercurial-scm.org/D9891

merge with stable

copyright: update to 2021 Differential Revision: https://phab.mercurial-scm.org/D9885

persistent-nodemap: Fix Rust declarations for Revlog_CAPI signatures Use Rust’s `libc::ssize_t` as the closest match to C’s `Py_ssize_t`. See details in test comment. Going forward we should find a way to have such Rust declarations auto-generated from C headers at build time, or auto-checked against them in a test. Differential Revision: https://phab.mercurial-scm.org/D9901

persistent-nodemap: add test case reproducing a Rust panic Thanks to Mitchell Plamann for reporting this bug. The next commit fixes it. Differential Revision: https://phab.mercurial-scm.org/D9900

copies: fix an incorrect comment in graftcopies() from recent D9802 I don't think the bit about creating a merge in the filelog was correct. Or at least I couldn't find a case where it happened. Differential Revision: https://phab.mercurial-scm.org/D9899

narrow: delete a stale TODO about not sending groups the client already has 2c5835b4246b changed the changegroup generation to not send treemanifests for directories the client had before widening. As that commit mentions, we had already stopped before that commit to send the changelog and filelogs for files the client already had. Differential Revision: https://phab.mercurial-scm.org/D9898

copies: add an devel option to trace all files Filelog based copy tracing only trace copy for file that have been added. This is a trade off between skipping some rare copy case in exchange for avoiding atrocious-to-the-point-of-unusable performance. The changeset centric copy tracing does not need this trade off and naturally trace all copy, include the one involving non-new files. In order to ease the comparison from both algorithm, we add a small devel option to trace copy for all files in the target revisions. Differential Revision: https://phab.mercurial-scm.org/D9796

revlog: prepare pure parser for being overloaded The current class uses module-level variables which don't allow for reusing the current logic for a slightly different revlog version. Differential Revision: https://phab.mercurial-scm.org/D9903