hook: move stdio redirection to context manager The old code was checking stdio redirection in a loop. This didn't make sense. The pattern is better expressed as a context manager IMO, so this commit refactors it to be one. Differential Revision: https://phab.mercurial-scm.org/D8338

pycompat: change argv conversion semantics Use of os.fsencode() to convert Python's sys.argv back to bytes was not correct because it isn't the logically inverse operation from what CPython was doing under the hood. This commit changes the logic for doing the str -> bytes conversion. This required a separate implementation for POSIX and Windows. The Windows behavior is arguably not ideal. The previous behavior on Windows was leading to failing tests, such as test-http-branchmap.t, which defines a utf-8 branch name via a command argument. Previously, Mercurial's argument parser looked to be receiving wchar_t bytes in some cases. After this commit, behavior on Windows is compatible with Python 2, where CPython did not implement `int wmain()` and Windows was performing a Unicode to ANSI conversion on the wchar_t native command line. Arguably better behavior on Windows would be for Mercurial to preserve the original Unicode sequence coming from Python and to wrap this in a bytes-like type so we can round trip safely. But, this would be new, backwards incompatible behavior. My goal for this commit was to converge Mercurial behavior on Python 3 on Windows to fix busted tests. And I believe I was successful, as this commit fixes 9 tests on my Windows machine and 14 tests in the AWS CI environment! Differential Revision: https://phab.mercurial-scm.org/D8337

phabricator: extract logic to print the status when posting a commit This will make it easier to list each commit when folding. That makes the output less confusing because it matches the output of `--confirm` and the revisions listed on the command line. Differential Revision: https://phab.mercurial-scm.org/D8313

phabricator: extract the logic to amend diff properties to a function This will be needed on a separate code path when dealing with folding revisions. And since we know that will involve adding multiple local commmits to the diff properties instead of just one, restructure the logic slightly to allow it. Differential Revision: https://phab.mercurial-scm.org/D8312

phabricator: teach `getoldnodedrevmap()` to handle folded reviews The tricky part here is reasoning through all of the possible predecessor scenarios. In the typical case of submitting a folded range and then resubmitting it (also folded), filtering the list of commits for the diff stored on Phabricator through the local predecessor list for each single node will result in the typical 1:1 mapping to the old node. There are edge cases like using `hg fold` within the range prior to resubmitting, that will result in mapping to multiple old nodes. In that case, the first direct predecessor is needed for the base of the diff, and the last direct predecessor is needed for the head of the diff in order to make sure that the entire range is included in the diff content. And none of this matters for commits in the middle of the range, as they are never used. Fortunately the only crucial thing here is the `drev` number for each node. For these complicated cases where there are multiple old nodes, simply ignore them all. This will cause `createdifferentialrevision()` to generate a new diff (within the same Differential), and avoids complicating the code. Differential Revision: https://phab.mercurial-scm.org/D8311

phabricator: teach createdifferentialrevision() to allow a folded commit range No visible changes here, until an option to enable it is added to `phabsend`. Differential Revision: https://phab.mercurial-scm.org/D8310

phabricator: combine commit messages into the review when folding commits No visible changes here, until an option to enable it is added to `phabsend`. This combines the Differential fields like Arcanist does, rather than simply concatenating the text blocks. Aside from populating everything properly in the web interface, Phabricator fails the review create/update if repeated fields are seen as would happen with simple concatenation. On the flip side, now that the Summary and Test Plan fields can contain data from multiple commits, we can't just join these fields together to determine if an amend is needed. If that were to happen, every single commit in the folded range would get amended with the combined commit message, which seems clearly wrong. Aside from making a minor assumption about the content of the Differential Revision field (it seems they allow some minor variances with spacing), this means that for folded reviews, you can't post it, go to the web page add a missing Test Plan, and then get it added to the commit message by re-posting it. I don't think that's a big deal. Differential Revision: https://phab.mercurial-scm.org/D8309

phabricator: record all local commits used to create a Differential revision Arcanist records all of the commits that it squashes into a single review, and that info will be helpful when adding similar functionality. This info is used when submitting an updated review, so that the extension can recalculate the old diff and see if a new one is necessary, or if it is just a property update. It also shows on the `commits` tab in the `Revision Contents` section. When submitting in the usual 1:1 commit to review mode, the wire protocol is unchanged. The content of `hg:meta` is a bit odd, but such is the problem when folding several commits. The choice for the parent node is obvious, but the `node` value uses the tip commit because that seems more natural, and is used elsewhere to look up the previous diff when updating. The rest of the attributes follow from there. Differential Revision: https://phab.mercurial-scm.org/D8308

tests: use `f --hexdump` to print file content The inline print.py in this test wasn't fully compatible with Python 3 because it was reading from sys.stdin, which already normalized line endings since it operates in the realm of str on Python 3. To do this correctly, we'd need to read from sys.stdin.buffer on Python 3. This would entail conditional code. I felt this was too much effort. So I just replaced the custom script with `f`, which already knows how to do the right thing. test-mactext.t now passes on Python 3 on Windows. Differential Revision: https://phab.mercurial-scm.org/D8336

url: pass str to pathname2url This is needed to appease Python 3. This fixes test-extdata.t and test-url-download.t on Python 3 on Windows. Differential Revision: https://phab.mercurial-scm.org/D8335

tests: pass str to matchoutput() It accepts a str, not bytes. This fixes a failure in test-hghave.t on Windows. Why it wasn't failing on Linux, I don't know. I suspect the Windows process code in Python doesn't accept bytes and the POSIX code does? Differential Revision: https://phab.mercurial-scm.org/D8334

hgcli: customize for Mercurial Now that we have a shiny new PyOxidizer-based hgcli project, let's customize it for Mercurial! This commit replaces the auto-generated pyoxidizer.bzl with one that installs Mercurial from the local source repository. A README.md with build instructions has been added. The Cargo.toml file has been updated to reflect the proper license and reference the added README.md. In my Linux environment, running the test suite yields 27 failures. It's worth noting the run time of the test harness on Linux on my Ryzen 3950X: before: 378s wall; 9982s user; 1195s sys after: 353s wall; 8996s user; 958s sys % orig: 93.4 wall; 90.1 user; 80.2 sys While I haven't measured explicitly, I suspect the performance win is due to in-memory resource loading (which is known to be faster than Python's filesystem importer). Differential Revision: https://phab.mercurial-scm.org/D8351

hgcli: add stub PyOxidizer project Using commit c772a1379c3026314eda1c8ea244b86c0658951d of PyOxidizer, I ran `pyoxidizer init-rust-project hgcli` to create a stub Rust project. The only modifications I made from what that command produced are: * Update location of pyembed crate to PyOxidizer's Git repository. * Removed some trailing whitespace from pyoxidizer.bzl * Added auto-generated Cargo.lock file Subsequent commits will modify the stub project to Mercurial's needs. Differential Revision: https://phab.mercurial-scm.org/D8350

hgcli: remove legacy project This code is a logical precursor to PyOxidizer. It is now defunct. Differential Revision: https://phab.mercurial-scm.org/D8349

nodemap: automatically "vacuum" the persistent nodemap when too sparse We arbitrarily pick "10%" as the threshold. Differential Revision: https://phab.mercurial-scm.org/D8193

nodemap: display percentage of unused in `hg debugnodemap` This is useful to assess the density of the cache. Differential Revision: https://phab.mercurial-scm.org/D8192

nodemap: make sure on disk change get rolled back with the transaction In case of errors, we need to rollback the change made to the persistent nodemap. Differential Revision: https://phab.mercurial-scm.org/D8191

nodemap: test that concurrent process don't see the pending transaction We don't want other client to read uncommitted data, until the transaction is really committed. Differential Revision: https://phab.mercurial-scm.org/D8190

testlib: adjust wait-on-file timeout according to the global test timeout Lets assume that if test timeout have been set to be twice as long, it means local timeout should be twice as long too. I am not aware of any case were extending timeout for file based synchronisation was necessary, but the safety seems simple to implements. Differential Revision: https://phab.mercurial-scm.org/D8316

testlib: add a small scrip to help process to synchronise using file Creating and waiting for files is a robust way to synchronise two processes running concurrently. We already use this approach in various tests. I am adding a official script to do so before adding more usage of this. Differential Revision: https://phab.mercurial-scm.org/D8189

setup: work around old versions of distutils breaking setup.py I'm not really sure how to trigger this, but we saw it in our build environment for Windows at Google. This fixed it. Sigh.

chgserver: update the umask cache before each run posix.py uses a global variable to store the umask value resulting in caching of it when using chg. We need to update it before each command run as the umask can change between commands. This fixes test-inherit-mode.t with chg.

tests: handle new error string from FreeBSD for dns entry not resolving Differential Revision: https://phab.mercurial-scm.org/D8333

notify: optional mail threading based on obsmarker When notify.reply is set and a changeset has a predecessor in the repository, include In-Reply-To pointing to the message-id that would have been generated for the oldest predecessor. This allows mail threading like Phabricator for common cases like rebasing changes, but will be optimal for cases like folding. Differential Revision: https://phab.mercurial-scm.org/D8172

pathutil: document that dirs map type implies manifest/dirstate processing

git: pass a list to pathutil.dirs to indicate that it is a manifest The python implementation of pathutil.dirs just uses a for loop which happens to work the same on both dicts and lists. The rust implementation actually figures out which of the two types it is, and directs the execution to either dirstate or manifest processing.

git: implement basic bookmark activation This is very limited, but it allows 'hg update foo' when already on foo. The caching is based on bmstore's caching.

git: implement a basic checkconflict bookmark store method It is heavily based on bmstore's own checkconflict.

git: abort when attempting to set a branch Given the mapping we use (namely, a git head is a bookmark), it is better to error out with a hint.

git: remove obsolete todo item The changes in 02c47b74366c cleaned up the requirement check.

tests: don't run test-update-atomic.t on chg I am unable to find a good way to make `showwrites` extension in it to work with chg. Also putting the use of showwrites inside `if no-chg` will defeat the purpose of test, so I just made the test no-op on chg.

tests: don't run couple of tests related to extensions loading with chg The couple of files test extensions loading and debugging that. It sets `devel.debug.extensions` to True to do that. Either we have to restart chg if this config is set or just don't run these tests on chg. I tried the first way by adding the config option to chgserver.py list of config subsections but that does not seem to work.

tests: conditionalize test-phases.t output for chg IIUC chg loads the ProgrammingError class which leads to it directly referencing it in error output. This makes the test pass on chg.

chg: be stricter about checking invocation of `serve` command Few tests run serve command in form of `hg -R <repo> serve` which leads to chg thinking that it's not a serve command and failing tests. We become more stricter in checking for the serve command. This fixes test-server-view.t, test-remote-hidden.t, test-remotefilelog-http.t, test-phases-exchange.t, test-wireproto-content-redirects.t with chg.

chgserver: add merge-tools to sensitive config items Because this can change whether the ui is gui or not. This fixes test-extdiff.t with chg.

run-tests: add --chg-debug flag to show chg debug output This has helped me a lot in debugging chg failures in tests.

tests: update test-ssh.t output with --chg The output change was caused by d7304434390f5efca405744fa12a6585edae3d83.

tests: update test-devel-warnings.t output with chg The output change was caused by dc9901558e3c6a78bad3f6594b3888f95104c443.

chgserver: add fastannotate config section to sensitive list Depending on the config value of `fastannotate.modes`, the fastannotate extension can do different things in uisetup. Depending on value of `fastannotate.server`, it can register new wireprotocol capabilities. This fixes test-fastannotate-hg.t, test-fastannotate-protocol.t and test-fastannotate.t with chg.

phabricator: account for `basectx != ctx` when calculating renames No functional changes here because the two are the currently same, but they won't be with a `--fold` option. Differential Revision: https://phab.mercurial-scm.org/D8307

phabricator: add basectx arguments to file related `phabsend` utilities This is in support of a future `--fold` option, that allows rolling up several commits into a single review with a diff from the start to the end of the range. There are no functional changes yet- the original `ctx` is also passed as the new `basectx`, which represents the first commit in the review range (similar to `qbase` in MQ parlance). Other functions will need the range of commits, but these deal with status or the diffs, so they only need the end points. Differential Revision: https://phab.mercurial-scm.org/D8306

phabricator: eliminate a couple of duplicate filectx lookups Differential Revision: https://phab.mercurial-scm.org/D8305

tests: update test-debugcommands.t stack trace checks for python3.9 Python 3.9 prints a complete filepath in more cases, so we need to handle that in our test output. Since we don't really care *that* much about the specific path formatting here, just use globbing. Differential Revision: https://phab.mercurial-scm.org/D8329

tests: update test-archive.t expectations for python3.9 changes Sigh. Differential Revision: https://phab.mercurial-scm.org/D8328

setup: relax -Werror for declaration-after-statement on Python 3.9 It turns out Python 3.9 introduces such declarations in the headers, eg cpython/abstract.h:189:5: error: ISO C90 forbids mixed declarations and code [-Werror=declaration-after-statement] so we have to be more relaxed when compiling for 3.9. Differential Revision: https://phab.mercurial-scm.org/D8327

pvec: drop an unused `from __future__ import division` This module only uses the `//` operator (for integer division). Differential Revision: https://phab.mercurial-scm.org/D8326

py3: use integer division in histedit Histedit uses the `/` operator, which does type conversion to float in Python 3 instead of integer division likeon Python 2. Let's preserve the Python 2 behavior by importing and using the `//` operator. Differential Revision: https://phab.mercurial-scm.org/D8324

rust: update README to reflect use of `regex` crate Differential Revision: https://phab.mercurial-scm.org/D8325

rust: update all dependencies We do this periodically to say up to date. No major versions were crossed this time per se, but the `rand` is still in v0, and their 0.7x series broke three things: - Some distribution-related elements were moved to a separate crate, flashing a deprecation warning - The `LogNormal::new` associated function now returns a `Result` - Certain RNGs were updated to sample a `u32` instead of `usize` when their upper-bound is less than `u32::MAX` for better portability, which changed the output for 2 tests. Moreover, the recent use of the `regex` crate for ignore mechanisms prompted some benchmarking that revealed that `regex` was slower at compiling big regex than `Re2`. The author of `regex` was very quick to discover an optimization that yielded a 30% improvement. It's still slower than `Re2` in that regard, but less so in the 1.3.6 release. Differential Revision: https://phab.mercurial-scm.org/D8320

rust: update micro-timer dependency The new version uses a much more robust technique and should remove any existing risk of bad compiler error or performance hit. Differential Revision: https://phab.mercurial-scm.org/D8319

rust-status: only involve ignore mechanism when needed This prevents unnecessary fallbacks to Python, improving performance for `hg update` for instance. On Mozilla-Central a noop update goes from 1.6s down to 700ms. Differential Revision: https://phab.mercurial-scm.org/D8315

templater: fix cbor() filter to recursively convert smartset to list The previous attempt, e3e44e6e7245 "templater: fix cbor() filter to accept smartset", was incomplete since obj may be a collection containing a smartset. This works around the problem by converting smartsets recursively. Another option is to teach cborutil how to encode a smartset. That should be okay, but I hesitated to add "import smartset" to cborutil.py as the cborutil is pretty generic.

shelve: split up dounshelve() in unshelvecmd() and _dounshelve() I'd like to be able to override the new `_dounshelve()`, getting access to the name of the shelve to unshelve. `unshelvecmd()` seems to better match the existing `createcmd()`, `listcmd()` etc. Differential Revision: https://phab.mercurial-scm.org/D8322

rust-matchers: use the `regex` crate Instead of falling back to Python when a code path with "ignore" functionality is reached and `Re2` is not installed, the default compilation (i.e. without the `with-re2` feature) will use the `regex` crate for all regular expressions business. As with the introduction of `Re2` in a previous series, this yields a big performance boost compared to the Python + C code in `status`, `diff`, `commit`, `update`, and maybe others. For now `Re2` looks to be faster at compiling the DFA (1.5ms vs 5ms for Netbeans' `.hgignore`) and a bit faster in actual use: (123ms vs 137ms for the parallel traversal of Netbeans' clean repo). I am in talks with the author of `regex` to see whether that performance difference is a bug, a "won't fix", or a tuning issue. The `regex` crate is already one of our dependencies and using this code does not require any additional work from the end-user than to use the Rust extensions. Differential Revision: https://phab.mercurial-scm.org/D8323

templater: add subsetparents(rev, revset) function Naming suggestions are welcome. And this could be flagged as an (ADVANCED) function since the primary use case is to draw a graph. This provides all data needed for drawing revisions graph filtered by revset, and allows us to implement a GUI graph viewer in some languages better than Python. A frontend grapher will be quite similar to our graphmod since subsetparents() just returns parent-child relations in the filtered sub graph. Frontend example: https://hg.sr.ht/~yuja/hgv/browse/default/core/hgchangesetgrapher.cpp However, the resulting graph will be simpler than the one "hg log -G" would generate because redundant edges are eliminated. This should be the same graph rendering strategy as TortoiseHg. This function could be implemented as a revset predicate, but that would mean the scanning state couldn't be cached and thus slow. Test cases are split to new file since test-template-functions.t is quite big and we'll need a new branchy repository anyway.

templater: remember cache key of evaluated revset This provides a hint for caching further computation result of the given revset. See the next patch for example.

templater: fix cbor() filter to accept smartset So the wrapper type can return a bare smartset.

templater: introduce wrapper for smartset (API) I want to add a template function which takes a revset as an argument: {somefunc(..., revset(...))} ^^^^^^^^^^^ evaluates to a revslist This wrapper will provide a method to get an underlying smartset. It should also be good for performance since count(revset(...)) will no longer have to fully consume the smartset for example, but that isn't the point of this change.

setup: build C extensions with -Werror=declaration-after-statement MSVC 2008 still needs declarations at the top of the scope. I added it to the 3rd party code too in case somebody vendors a new version with a problem- they'll get an early warning. Clang seems to ignore this (at least on 10.14 with Xcode 10), and gcc 7.4 will error out as desired on Ubuntu 18.04. Thanks to Yuya for remembering the name of the option. Differential Revision: https://phab.mercurial-scm.org/D8318

merge with stable