rust-discovery: core implementation for take_quick_sample() This makes in particular `rand` no longer a testing dependency. We keep a seedable random generator on the `PartialDiscovery` object itself, to avoid lengthy initialization. In take_quick_sample() itself, we had to avoid keeping the reference to `self.undecided` to cope with the mutable reference introduced by the the call to `limit_sample`, but it's still manageable without resorting to inner mutability. Sampling being prone to be improved in the mid-term future, testing is minimal, amounting to checking which code path got executed. Differential Revision: https://phab.mercurial-scm.org/D6423

rust-discovery: read the index from a repo passed at init This makes the API of the Rust PartialDiscovery object now the same (or rather a subset) of the Python object, hence easier to control through module policy down the road. Differential Revision: https://phab.mercurial-scm.org/D6517

rust-discovery: accept the new 'respectsize' init arg At this stage, we don't do anything about it: it will be meaningful in sampling methods that aren't implemented yet. Differential Revision: https://phab.mercurial-scm.org/D6516

merge with stable

unshelve: forget unknown files after a partial unshelve This is a follow-up patch to 6957f7b93e03. This allows hg to forget unknown files after a partial unshelve. Differential Revision: https://phab.mercurial-scm.org/D6724

flagutil: move addflagprocessor to the new module (API)

flagutil: move insertflagprocessor to the new module (API)

flagutil: move REVIDX_KNOWN_FLAGS source of truth in flagutil (API) Since REVIDX_KNOWN_FLAGS is "not really a constant" (extension can update it) and python integer,... it needs to be the responsability of a single module and always accessed through the module. We update all the user and move the source of truth in flagutil.

flagutil: move the `flagprocessors` mapping in the new module This module is meant to host most of the flag processing logic. We start with the mapping between flag and processors.

flagutil: create a `mercurial.revlogutils.flagutil` module The flagprocessings logic is duplicated in 2 extra places, and usually in a less robust flavor. This is a maintenance nightmare that I would like to see cleaned up. To do so I am creating a `flagutil` module to move flag processings related code and make it easily reusable by other code.

rawdata: register the method for `ifiledata` The interface have a `revision(..., raw=False)` method so it should get a `rawdata` one. I am not sure why nothing complained about the lack of it earlier.

rawdata: implement the method for `unionrepo` too This is required for all implementations.

rawdata: implement the method for `remotefilelog` too This is needed for all storage implementations.

rawdata: implement `rawdata` for `simplestore` too This is needed for all implementation.

rawdata: forward `rawdata` call on `manifestlog` This needs to be sent to the underlying `revlog` too.

rawdata: implement `rawdata` for `sqlitestore` too This is a different store, it needs it declared.

rawdata: add the method to bundlerevlog The bundlerepo logic has its own `revision` method on its own `revlog` object. We need to "implement" `rawdata` there too.

rawdata: forward the method call on `filelog` object We have a new method, we need to expose it.

rawdata: introduce a `rawdata` method on revlog This method aims at replacing `revision(..., raw=True)` call. The purpose of data returned without and without raw are different enough that having two different method would make sense. This split is motivated by other work aiming at storing data on the side of the main revision of a revlog. Having a cleaner API makes it simpler to add this work. The series following this first changesets is organised as follow: 1) add `rawdata` method everywhere it is useful 2) update all caller 3) implement all `rawdata` method without using `revision` 4) deprecate the `rawdata` parameter of `revision`

revlog: split a `_revisiondata` method to file `revision` job We are about to introduce more public method to access revision data (eg: `rawdata`). revset subclass tend to recursively call `revision` which will create all kind of issue with the coming series. To avoid them we introduce an explicit difference between the internal call and the public all. This will be useful for later work anyway (so the subclass issue is just moving it earlier in the series). I am not sure if the subclass are actually doing something sensible. However, I am certain I don't want to be rabbit holed into figuring it out right now.

continue: added support for transplant This creates a seperate function `continuetransplant()` containing logic for resuming transplant from interrupted state. `continuetransplant()` is then registered as `continuefunc` for state detection API. Results are shown in tests. Differential Revision: https://phab.mercurial-scm.org/D6689

merge with stable

branchmap: explicitly warm+write all subsets of the branchmap caches 'full' claims it will warm all of the caches that are known about, but this was not the case - it did not actually warm the branchmap caches for subsets that we haven't requested, or for subsets that are still considered "valid". By explicitly writing them to disk, we can force the subsets for ex: "served" to be written ("immutable" and "base"), making it cheaper to calculate "served" the next time it needs to be updated. Differential Revision: https://phab.mercurial-scm.org/D6710

changectx: extract explicit computechangesetfilesremoved method from context Right now, the logic around changeset centric removed files data are buried into the "changectx" code. We extract this code in a dedicated method (in the scmutil module) for clarity. This clarity will help to explicitly compute and caches these data in the future.

changectx: extract explicit computechangesetfilesadded method from context Right now, the logic around changeset centric added files data are buried into the "changectx" code. We extract this code in a dedicated method (in the scmutil module) for clarity. This clarity will help to explicitly compute and caches these data in the future.

copies: extract an explicit `computechangesetcopie` method from context Right now, the logic around changeset centric copies data are buried into the "changectx" code. We extract this code in a dedicated method (in the copies module) for clarity. This clarity will help to explicitly compute and caches these data in the future.

config: fix fm.data() handling of defaultvalue This is a follow-up patch to rHG51a2e3102db2. This moves `fm.data()` out of the if block in `commands.config()`. Differential Revision: https://phab.mercurial-scm.org/D6720

config: remove pycompat.bytestr() for defaultvalue This is a follow-up patch to 51a2e3102db2. This removes `pycompat.bytestr` to preserve `None` in `commands.config()`. Differential Revision: https://phab.mercurial-scm.org/D6712

unshelve: clear shelvedstate and _finishunshelve() on partial unshelve On a partial unshelve, `shelvedstate` was not cleared and `_finishunshelve()` was not called. Ideally, these should be called on this case. This patch makes `shelvedstate` to delete after a successful partial unshelve and calls `_finishunshelve()` in the same case. Differential Revision: https://phab.mercurial-scm.org/D6708

unshelve: delete shelvedstate after a successful unshelve --continue `unshelve --continue` was preventing the deletion of `shelvedstate` on a partial `unshelve`. Ideally, `shelvedstate` should be deleted after a successful `unshelve`. Now, the behavior of `unshelve --continue` will be as follows in interactive mode: 1] The user tried to `unshelve` changes interactively but, ran into conflicts. 2] They resolved the conflicts and triggered `unshelve --continue` but, unshelved changes partially. 3] Now, on trying to do `unshelve --continue` again will abort as the last `unshelve` was successful and we are deleting the `shelvedstate`. 4] If they want to unshelve the remaining shelved change, they need to trigger `unshelve` without `--continue`. Differential Revision: https://phab.mercurial-scm.org/D6694

unshelve: handle stripping changesets on interactive mode On interactive mode, changesets on `nodestoremove` should be stripped regardless of the shelve is partial or not. This patch modifies `unshelvecontinue()` to do that. Differential Revision: https://phab.mercurial-scm.org/D6686

byteify-strings: add --version argument This is indispensable for automated tools to detect changes in behavior.

byteify-strings: add space in special comments to silence flake8 error This is done soon enough that nobody has had the time to use this feature yet.

revset: drop argument when it's None getstack's definition is `getstack(repo, rev=None)`, so providing None explicitly is unnecessary. Moreover, when x is not None, it's definitely not a revision but a part of a parsed tree of revset arguments. Differential Revision: https://phab.mercurial-scm.org/D6707

stack: remove unnecessary reverse() predicate Stack already sorts revisions, so no need to do it twice. This change was a part of D2400, which didn't land for other reasons. See also D2399, where this change was suggested. Differential Revision: https://phab.mercurial-scm.org/D6706

automation: increase root volume size on Linux It is close to full in the AMI. I actually ran out of space running tests without increasing the volume size! Differential Revision: https://phab.mercurial-scm.org/D6716

automation: install Rust in Linux environment This will install Rust 1.31.1, 1.34.2, and whatever stable is at the time the install runs. We install 1.31.1 as our minimum supported Rust version (I think that's what we're currently targeting) and 1.34 because that's what Debian 10 is shipping. Differential Revision: https://phab.mercurial-scm.org/D6715

automation: update packages in requirements files We like keeping up to date. The content of the autogenerated files changed slightly because I used a newer version of `pip-compile` than what was used previously. Differential Revision: https://phab.mercurial-scm.org/D6714

automation: install latest Python versions This required bumping the pyenv commit so the new versions are available. Differential Revision: https://phab.mercurial-scm.org/D6713

upgrade: introduce the internal code for revlog cloning selection For now we still clone every single revlogs but all the selection mechanism is now in place in the lower layer. The next changesets will introduce the user interface part of the selection.

upgrade: introduce a _copyrevlog method This function copies a revlog from the old store to the new one, without re-adding all deltas manually. This will eventually save a lot of time when some revlog does not needs any conversions. Code actually using this will be introduced in later changesets.

upgrade: rename `_copyrevlogs` to `_clonerevlogs` The underlying revlog method is named `clone`, keeping the naming consistent seems clearer. This is motivated to clarify the difference with an (upcoming) function that simply copy revlog files as is.

upgrade: walk the source store file only once I don't expect this to have a significant performance impact, but it seems simpler and saner to do the operation only once and to keep the result around.

upgrade: always use full text if "full-add" mode is enable We should not be using a delta since the goal is to perform a full addition from scratch in all cases. Without this patch, `hg debugupgraderepo --optimize re-delta-fulladd --run` can crash.

byteify-strings: fix misalignment with multi-line parenthesis This improves the current fix to also take into account cases where the last line ended on the opening `(`, `[` or `{` and adds a regression test.

byteify-strings: add test for byteify-strings.py This tests the basic features expected from this script, some cases may not be covered yet. A future patch will demonstrate an issue with multi-line `(`, `[` and `{` alignment and propose a fix.

merge with stable

byteify-strings: add cli argument to handle `attr*()` when they are methods Certain code bases have useful utils that wrap the builtin functions, and are called like `util.setattr`.

byteify-strings: simplify default value for `--treat-as-kwargs`

byteify-strings: add --treat-as-kwargs argument to handle kwargs-like objects This argument will help extensions move to Python 3 as keyword arguments should not be byte-prefixed. Most of the time, code bases will call this object `kwargs`, but other conventions exist like `opts`, so it should make sense to allow for custom names. This is a best effort solution that does minimal static checking; cases like `options = [o for o in ('a', 'b', 'c') if kwargs.get(o)]` and other just as complicated will not be detected.

byteify-strings: add helpers to check for item access or method call These helpers will be used in a future patch, split for ease of review.

byteify-strings: add support for ignore comments Our simple token analysis is sometimes not clever enough, we need to be able to turn off our script for parts of the code. This change introduces three special comments: - `#no-py3-transform` to tell `byteify-strings` ignore the next line - `#py3-transform: off` to ignore everything until the end of the file - `#py3-transform: on` to stop ignoring The last two can be particularly useful within Python 2/3 compatibility files.

byteify-strings: handle triple quoted strings if they are not docstrings As with anything in this script, this is a best effort approach. Most of the time, when a triple quoted string is assigned to something, it's not a docstring.

byteify-strings: handle multi-line strings in _ensuresysstr The current implementation did not handle calls like `repo.ui.log("first line" "other line")` correctly.

fix: run fixer tools in the repo root as cwd so they can use the working copy This lets fixer tools do things like find configuration files, with the caveat that they'll only see the version of that file in the working copy, regardless of what revisions are being fixed. Differential Revision: https://phab.mercurial-scm.org/D6440

config: add defaultvalue template keyword This patch tries to fix one of the issues mentioned in issue6014. This adds a new `defaultvalue` template keyword to be used with `hg showconfig` to get the default value of the config item. Differential Revision: https://phab.mercurial-scm.org/D6704

merge with stable

module-policy: update rust extension import to use the new module policy Differential Revision: https://phab.mercurial-scm.org/D6677

transaction: leave unfinished without crashing when not properly released I think the transaction.__del__ is there just as a last resort in case we (or an extension) forgot to release the transaction. When that happens, the repo can (or will on Python 3?) get deleted before the transaction. This leads to a crash in test-devel-warnings.t on Python 3 because we tried to access repo.dirstate, where repo was retried from a weak reference. There's not much we can do here, but let's at least avoid the crash. The user will have run `hg recover` afterwards regardless. Differential Revision: https://phab.mercurial-scm.org/D6664

unshelve: add abort on using continue and interactive together `unshelve --continue --interactive` will not work as expected by the user as the mode of in-progress unshelve is preserved and cannot be overwritten. This patch makes `unshelve` to throw an error on using both `--continue` and `--interactive` together with `unshelve`. Differential Revision: https://phab.mercurial-scm.org/D6703