revlog: add opener option to enable ellipsis flag processor The ellipsis flag processor can now be registered by specifying an opener option when constructing a revlog instance. This allows us to enable ellipsis flags on a per-revlog basis. Differential Revision: https://phab.mercurial-scm.org/D4647

revlog: store flag processors per revlog Previously, revlog flag processing would consult a global dict when processing flags. This was simple. But it had the undesired side-effect that any extension could load flag processors once and those flag processors would be available to any revlog that was subsequent loaded in the process. e.g. in hgweb, if the narrow extension were loaded for repo A but not repo B, repo B would be able to decode ellipsis flags even though it shouldn't be able to. Making the flag processors dict per-revlog allows us to have per-revlog controls over what flag processors are available, thus preserving desired granular access to flag processors depending on the revlog's needs. If a flag processor is globally registered, it is still globally available. So this commit should not meaningfully change behavior. Differential Revision: https://phab.mercurial-scm.org/D4646

revlog: define ellipsis flag processors in core We will soon be teaching core to honor the ellipsis flag on revlogs. Moving the definition of the processor functions to core is the first step in this. The processor is still not registered unless the narrow extension is loaded. Differential Revision: https://phab.mercurial-scm.org/D4645

narrow: remove custom filelog type This functionality is now handled by core as of the previous commit. I wanted this to be a standalone commit because the deleted code makes a reference to remotefilelog's file type missing a node() method and this may have implications to narrow+remotefilelog usage. The code in core doesn't perform this check and therefore behavior may be subtly different and buggy. But I /think/ the check is merely a performance optimization and nothing more. So I'm optimistic this will continue to "just work." Differential Revision: https://phab.mercurial-scm.org/D4644

filelog: custom filelog to be used with narrow repos Narrow repos may have file revisions whose copy/rename metadata references files not in the store. This can pose problems when consumers attempt to access a missing referenced file revision. The narrow extension hacks around this problem by implementing a derived filelog type that provides custom implementations of renamed(), size(), and cmp() which handle renames against files not in the narrow spec by silently removing the rename metadata. While silently dropping metadata isn't the most robust solution, it is the easiest to implement. This commit ports the custom narrow filelog class to core. When a narrow repo is constructed, its ifilestorage creation function will automatically use the new filelog type. This means the extra logic is 0 cost for non-narrow repos and shouldn't interfere with their operation. Differential Revision: https://phab.mercurial-scm.org/D4643

localrepo: iteratively derive local repository type This commit implements the dynamic local repository type derivation that was explained in the recent commit bfeab472e3c0 "localrepo: create new function for instantiating a local repo object." Instead of a static localrepository class/type which must be customized after construction, we now dynamically construct a type by building up base classes/types to represent specific repository interfaces. Conceptually, the end state is similar to what was happening when various extensions would monkeypatch the __class__ of newly-constructed repo instances. However, the approach is inverted. Instead of making the instance then customizing it, we do the customization up front by influencing the behavior of the type then we instantiate that custom type. This approach gives us much more flexibility. For example, we can use completely separate classes for implementing different aspects of the repository. For example, we could have one class representing revlog-based file storage and another representing non-revlog based file storage. When then choose which implementation to use based on the presence of repo requirements. A concern with this approach is that it creates a lot more types and complexity and that complexity adds overhead. Yes, it is true that this approach will result in more types being created. Yes, this is more complicated than traditional "instantiate a static type." However, I believe the alternatives to supporting alternate storage backends are just as complicated. (Before I arrived at this solution, I had patches storing factory functions on local repo instances for e.g. constructing a file storage instance. We ended up having a handful of these. And this was logically identical to assigning custom methods. Since we were logically changing the type of the instance, I figured it would be better to just use specialized types instead of introducing levels of abstraction at run-time.) On the performance front, I don't believe that having N base classes has any significant performance overhead compared to just a single base class. Intuition says that Python will need to iterate the base classes to find an attribute. However, CPython caches method lookups: as long as the __class__ or MRO isn't changing, method attribute lookup should be constant time after first access. And non-method attributes are stored in __dict__, of which there is only 1 per object, so the number of base classes for __dict__ is irrelevant. Anyway, this commit splits up the monolithic completelocalrepository interface into sub-interfaces: 1 for file storage and 1 representing everything else. We've taught ``makelocalrepository()`` to call a series of factory functions which will produce types implementing specific interfaces. It then calls type() to create a new type from the built-up list of base types. This commit should be considered a start and not the end state. I suspect we'll hit a number of problems as we start to implement alternate storage backends: * Passing custom arguments to __init__ and setting custom attributes on __dict__. * Customizing the set of interfaces that are needed. e.g. the "readonly" intent could translate to not requesting an interface providing methods related to writing. * More ergonomic way for extensions to insert themselves so their callbacks aren't unconditionally called. * Wanting to modify vfs instances, other arguments passed to __init__. That being said, this code is usable in its current state and I'm convinced future commits will demonstrate the value in this approach. Differential Revision: https://phab.mercurial-scm.org/D4642

localrepo: pass root manifest into manifestlog.__init__ Today, localrepository has a method that can be overloaded which returns an instance of the root manifest storage object. When a manifestlog is created, it calls this private method and stores the root manifest object on it. This "hook" on localrepository isn't part of the documented interface. It isn't compatible with our desire to make repo storage determined before the repo object is constructed. This commit changes manifestlog.__init__ to accept the root storage object instead of calling into the repo to construct it. By doing things this way, the repo instance is responsible for constructing the manifest storage object directly. This does mean that other derived repo types need to overload manifestlog(). But they should have been doing this already, as manifestlog() is typically decorated in a storage-specific way. e.g. localrepository.manifestlog() is decorated as @storecache('00manifest.i'). And this assumes that a 00manifest.i file exists in the store vfs. This condition may not hold for repository types using non-revlog storage. So it is important for special repo types to override manifestlog() to remove this file association. The code changed in perf is wrong because it isn't compatible with older Mercurial versions. But I'm pretty sure the code was broken on older versions before this commit. It only affects `hg perftags`. I don't care enough to fix that at this time. .. api:: ``manifest.manifestlog.__init__()`` now receives the root manifest storage instance instead of calling into a private method on the repo object to obtain it. Differential Revision: https://phab.mercurial-scm.org/D4641

py3: create built in exceptions with str type messages in win32.py I hit an IOError in unlink() in test-pathconflicts-basic.t, that then crashed as it was handled: File "mercurial\dispatch.py", line 359, in _runcatch return _callcatch(ui, _runcatchfunc) File "mercurial\dispatch.py", line 367, in _callcatch return scmutil.callcatch(ui, func) File "mercurial\scmutil.py", line 252, in callcatch ui.error(_("abort: %s\n") % encoding.strtolocal(inst.strerror)) File "mercurial\encoding.py", line 205, in unitolocal return tolocal(u.encode('utf-8')) AttributeError: 'bytes' object has no attribute 'encode'

tests: stabilize test-shelve.t#phasebased for #no-symlink and #no-execbit The rev number ended up being 11 instead of 13 on Windows. If I ever get back to issue2020, this will go away.

debugdirstate: deprecate --nodates in favor of --no-dates We have supported 'no-' prefixes for boolean flag for a few years now, so I was expecting it to be --no-dates. I noticed that we have --nodates options for a few more commands (e.g. `hg diff`), but I'll leave that for another day. Differential Revision: https://phab.mercurial-scm.org/D4693

py3: fix a type error in hghave.has_hardlink test-hghave.t was failing with: feature hardlink failed: argument 1: <class 'TypeError'>: wrong type

narrow: remove hack to read narowspec from shared .hg directory This was another leftover from 576eef1ab43d (narrow: move .hg/narrowspec to .hg/store/narrowspec (BC), 2018-08-02), in addition to 623081f2abc2 (narrow: remove hack to write narrowspec to shared .hg directory, 2018-09-12). Differential Revision: https://phab.mercurial-scm.org/D4692

streamclone: reimplement nested context manager It's gone in Python 3, and you can't *ctxs into a with statement. Sigh. Differential Revision: https://phab.mercurial-scm.org/D4690

bundle2: grab kwarg using sysstr # skip-blame just an r prefix on a string Differential Revision: https://phab.mercurial-scm.org/D4691

py3: mark another passing test Differential Revision: https://phab.mercurial-scm.org/D4689

bookmarks: remove --active in favor of --list It's weird that we have both --active and --inactive options meaning completely different things. Instead of adding a one-off option, let's document the way to display the active bookmark by using -l/--list. No deprecated option is added since --active isn't released yet.

bookmarks: add explicit option to list bookmarks of the given names This is a generalized form of the --active option. A redundant sorted() call is removed. There was no point to update dict items in lexicographical order.

bookmarks: reject --delete with --inactive which makes no sense A deleted bookmark is neither active nor inactive.

bookmarks: parse out --inactive to action early The --inactive option can't be directly mapped to an action or a modifier. With any names, it means to add/rename to inactive bookmarks. Without names, it means to deactivate the current bookmark. This patch separates them to "inactive" flag and "action == 'inactive'".

bookmarks: parse out implicit "add" action early This prepares for adding -l/--list option, which can be combined with the positional arguments.

bookmarks: clarify that opts['rename'] points to an old bookmark to be renamed

bookmarks: refactor option checking to pick one from --delete/rename/active

bookmarks: convert opts to bytes dict early

bookmarks: pass in formatter to printbookmarks() instead of opts (API) This clarifies that user options have to be processed before calling printbookmarks().

strip: ignore orphaned internal changesets while computing safe strip roots Internal changeset can be safely garbage collected, so we can ignore them during safestrip. (Another phase for internal changeset that must be kept in the repository might be introduced later).

shelve: no longer strip internal commit when using internal phase When the internal phase is used, the internal commits we create during shelve will be automatically hidden, and we don't need to strip them. Avoiding strips gives much better performances and is less traumatic for caches. Test changes are all related to revision numbers increasing more quickly since we avoid stripping. At the end of `test-shelve.t` we now need manually strip the shelve-commit in addition to the x.shelve file deletion. This emulates a preexisting shelve after a repository upgrade. Note: The hidden internal commits confuses rebase a bit as shown by a new test added. This will happen when the user have shelve commits on top of a changeset to be rebased. We'll fix this in the next commit. As we still use a backup bundle, rebase can just strip the internal changesets and be fine.

meld: enable auto-merge This tells meld to resolve trivial conflicts before presenting the user with the remaining conflicts. This was attempted 5 years ago, but then --auto-merge was too new that the patch was rejected out of concern that users still had an older version of meld installed [1]. Maybe it's safe to assume that they have a newer version now. [1] https://www.mercurial-scm.org/pipermail/mercurial-devel/2013-April/050084.html Differential Revision: https://phab.mercurial-scm.org/D4665

run-tests: partially backout PYTHON quoting In 7f8b7a060584, I quoted this to support python being installed to "Program Files". Even though the string passed to os.popen() is this: "c:/Python27/python.exe" -c "import mercurial; print (mercurial.__path__[0])" ... cmd.exe is trying to run this: 'c:/Python27/python.exe" -c "import' This caused test-hghave.t to fail, reporting 'unexpected mercurial lib: ""', because the failed execution prints nothing to stdout. Py3 fails as though it's not quoted. For whatever reason, print() shows up in the output when run with py2, but not py3, so I'm having a hard time debugging this. For now, let's fix the buildbot.

py3: use '%d' instead of '%s' for integers Python 3 does not allow to use '%s' for integers. Differential Revision: https://phab.mercurial-scm.org/D4688

py3: use print as a function in tests/test-revert.t This makes the test work on Python 3. Differential Revision: https://phab.mercurial-scm.org/D4687

chgserver: restore pager fds attached within runcommand session While rewriting chg in Rust, I noticed the server leaks the client's pager fd. This isn't a problem right now since the IPC process terminates earlier than the pager, but I believe the fds attached within a "runcommand" request should be released as soon as the session ends.

chgserver: add separate flag to remember if stdio fds are replaced I want to make it use a separate saved buffer for "attachio" requests within "runcommand" session. See the next patch for details.

status: remove "morestatus" message from formatter data (BC) They are just printable messages, not data that should be fed to JSON or templater.

tests: show that the structure of the more status output looks weird Each dict should represent data of the same kind.

phabricator: add testedwith boilerplate

narrow: extract wdir cleanup function to make it extensible We have an overlay filesystem which shows the entire repository, and unlinking a file that's in the underlying data store will create "tombstone" entries, which are going to cause our automatic tracking to re-add these directories. We need to use a different (non-posix) interface to clean up items in the working directory that are no longer relevant. Extracting this to a function lets us use extensions.wrappedfunction and perform this cleanup work, even if the paths aren't in the dirstate (they may have been removed in the past and thus entirely "tombstone" entries already, part of hgignore, exclusively directories (possibly empty), or other edge cases). Differential Revision: https://phab.mercurial-scm.org/D4681

changegroup: reintroduce some comments that have gotten lost over the years I got concerned about the correctness of the pruning logic, but I was misreading it. I didn't figure that out until I walked all the way back to 0252abaafb8a from 20111, where I was finally able to see (in the deleted side of the change!) a complete explanation from b6d9ea0bc107 in 2005. Differential Revision: https://phab.mercurial-scm.org/D4686

changegroup: tease out a temporary prune method for manifests It's extracted so extensions can filter manifest nodes if needed. This is an unfortunate hack, but I think I only need it for manifests. The long-term solution will be to rework the relationship between changegroups and storage so that this isn't required. Differential Revision: https://phab.mercurial-scm.org/D4685

changegroup: remove outdated comment Differential Revision: https://phab.mercurial-scm.org/D4684

py3: encode the name to bytes before using in revsetpredicate() Differential Revision: https://phab.mercurial-scm.org/D4677

py3: suppress the output on .write() calls in tests/test-hgweb-commands.t Differential Revision: https://phab.mercurial-scm.org/D4676

py3: use stringutil.pprint() to print boolean values Differential Revision: https://phab.mercurial-scm.org/D4675

py3: add a missing b'' in tests/test-newercgi.t # skip-blame because just b'' prefixes Differential Revision: https://phab.mercurial-scm.org/D4674

py3: use pycompat.maplist instead of map Differential Revision: https://phab.mercurial-scm.org/D4673

py3: add some b'' prefixes in tests/test-extension.t # skip-blame because just b'' prefixes Differential Revision: https://phab.mercurial-scm.org/D4672

py3: make tests/svn-safe-append.py compatible with python 3 Differential Revision: https://phab.mercurial-scm.org/D4671

py3: use print as a function in tests/test-subrepo-svn.t Differential Revision: https://phab.mercurial-scm.org/D4670

bundle2: make server.bundle2.stream default to True Support for bundle2 streaming clones has been shipped in Mercurial 4.5 (7eedbd5d4880), but was never activated by default. It's time to have more people use it. The new format allows streaming clones to transport cache (hooray for speed) and phaseroots (fixes phase-related issues). Changes in tests: bundle2 capabilities now have "stream=v2" (plus a '\n' as a separator) and therefore take 14 bytes more: "%0Astream%3Dv2". Tip for tests that have data encoded with CBOR: 0xd3 - 0xc5 = 14. $USUAL_BUNDLE2_CAPS$ replaces $USUAL_BUNDLE2_CAPS_SERVER$, which is the same thing, but without "stream=v2". Since streaming clones now also transfer caches, the reported byte and file counts are higher (e.g. 816 bytes in 9 files instead of 613 bytes in 4 files, a bit of --debug and manual math confirms that the caches take these extra 203 bytes in 5 files). Differential Revision: https://phab.mercurial-scm.org/D4680

bundle2: graduate bundle2.stream option from experimental to server section Differential Revision: https://phab.mercurial-scm.org/D4679

tests: split capabilities into separate lines while searching for "narrow" This test is interested only in capabilities that are related to narrow, so let's omit everything else. Makes it easier to update other capabilities (and "rev-branch-cache" is one of the usual patterns that are already present in tests/common-patterns.py anyway). Differential Revision: https://phab.mercurial-scm.org/D4678

py3: resolve Unicode issues around `hg serve` on Windows Presumably we're going to want to use CreateProcessW(), and possibly get rid of pycompat.getcwd() here (which maps to the DeprecationWarning causing os.getcwdb()) to use os.getcwd() directly. But this was a minimal change to get rid of some stacktraces in test-run-tests.t.

run-tests: avoid os.getcwdb() on Windows Any call to this issues a DeprecationWarning about the Windows bytes API being deprecated. There are a handful of these calls in core, but test-run-tests.t was littered with these, as it's printed everytime run-tests.py is launched. I'm not sure what the long term strategy for Unicode on Windows in the test runner is, but this seems no worse than the current conversion strategy.

run-tests: quote PYTHON when spawning a subprocess Same reason as 5abc47d4ca6b. This covers running *.py tests, as well as inline python blocks. I didn't hit the path around line 3079, but it seems correct to quote.

narrow: add test showing that local-to-local narrow clones don't work It turns out they've never actually worked: prior to some recent refactoring they just unintentionally followed the full-clone path, which we unintentionally relied on in a test at Google. Differential Revision: https://phab.mercurial-scm.org/D4640

fastannotate: process files as they arrive peer.commandexecutor()'s context manager waits for all responses to arrive in its __exit__() method. We want to process the results as they arrive, so we should do that inside the context manager scope. Note that the futures' result() methods have been replaced to make sure that the command executor's sendcommands() method is called when the first future's result is requested, so we don't need to do that. A minor side-effect is that we can no longer easily tell when the server has started sending us responses, so that long statement was lost. Differential Revision: https://phab.mercurial-scm.org/D4666

py3: make osenvironb a proxy for, instead of a copy of os.environ where needed Without this, TESTDIR and a few other variables weren't defined in the *.t test. I didn't bother implementing all of the view functions for simplicity. All that is actually used is __{get,set}item__(), get() and pop(), but the rest seems easy enough to add for futureproofing.

memctx: simplify _manifest with new revlog nodeids This was originally written before we had modifiednodeid and addednodeid, so we had to get the parents of the context, the data from the function, and then hash that. This is much more simple now and helps refactor more code later.

context: remove unused overlayfilectx (API) It seems that this was maybe used in an extension but at this point nothing in lfs, hg-experimental, or any other cursory repo looked at has a reference to this class; so, for now, let's just remove it.

context: fix typo in workingcommitctx This was probably a copy pasta error in 745e3b485632. Refactoring memctx code exposed this bug.

filectx: fix return of renamed How is this not blowing up everywhere? It seems that filelog.renamed has always returned False (incorrectly a boolean) instead of the assumed None. Tracing through history, you need to skip over my move of code in 2013 by annotating from 896193a9cab4^ and you can see the original code is from 2007 (180a3eee4b75) and that ab9fa7a85dd9 broke this by assuming renamed was a bool (instead of None). Refactoring memctx code later exposed this bug.

tests: glob over some quoting differences in test-narrow-widen-no-ellipsis.t

py3: byteify contrib/check-config.py The corresponding *.t still fails because of bytes (with a 'b' prefix) vs str printing, but no longer crashes. # skip-blame for b'' prefixing

tests: quote PYTHON usage Python3 defaults to installing under "Program Files".

py3: add a missing b'' for Windows I tried ./contrib/byteify-strings.py, but there were way too many changes (and most looked wrong). This was hit with test-check-interfaces.py. # skip-blame for b'' prefixes

log: make changesetformatter pass in changectx to formatter It wasn't necessary before, but user templates may have keywords that aren't filled in by the changesetformatter.

journal: use changesetformatter to properly nest list of commits in JSON Before, two separate JSON documents were interleaved. I chose the field name "changesets" over the option name "commits", since each entry is called a "changeset" in log templates.

journal: do not pass in repolookuperror string to template (BC) This doesn't look like data, but a warning message.

journal: inline formatted nodes and date into expression The variable name "str" was misleading since these values aren't always strings.

journal: unify template name for "nodes" (BC) This is a part of the name unification. https://www.mercurial-scm.org/wiki/GenericTemplatingPlan#Dictionary .. bc:: ``{oldhashes}`` and ``{newhashes}`` in journal template are renamed to ``{oldnodes}`` and ``{newnodes}`` respectively.

localrepo: extract resolving of opener options to standalone functions Requirements and config options are converted into a dict which is available to the store vfs to consult. This is how storage options are communicated from the repo layer to the storage layer. Currently, we do that option resolution in a private method on the repo instance. And there is a single method doing that resolution. Opener options are logically specific to the storage backend they apply to. And, opener options may wish to influence how the repo object/type is constructed. So it makes sense to have more granular storage option resolution that occurs before the repo object is instantiated. This commit extracts the code for resolving opener options into new module-level functions. These functions are run before the repo instance is constructed. As part of the code move, we split the option resolution into generic and revlog-specific options. After this commit, we no longer add revlog-specific options to repos that don't have a revlog requirement. Some of these opener options and associated config options might make sense on alternate storage backends. We can always reuse config options and opener option names for other backends. But we shouldn't be passing opener options to storage backends that won't recognize them. I haven't done it here, but after this commit it should be possible for store backends to validate the set of opener options it receives. Because localrepository.openerreqs is no longer used after this commit, it has been removed. I'm not super thrilled about the code outside of localrepo that is adding requirements and updating opener options. We'll probably want to create a more formal API for that use case that constructs a new repo instance and poisons the old repo object. But this was a pre-existing issue and can be dealt with later. I have little doubt it will cause me troubles as I continue to refactor how repository objects are instantiated. .. api:: ``localrepository.openerreqs`` has been removed. Override ``localrepo.resolvestorevfsoptions()`` to add custom opener options. .. api:: ``localrepository._applyopenerreqs()`` has been removed. Use ``localrepo.resolvestorevfsoptions()`` to add custom opener options. Differential Revision: https://phab.mercurial-scm.org/D4576

localrepo: use boolean in opener options Not sure why we're using an integer for a flag value here. I'm pretty sure nothing relies on values being 1. While we're here, convert to a dict comprehension. Differential Revision: https://phab.mercurial-scm.org/D4575

localrepo: move store() from store module I want logic related to requirements handling to be in the localrepo module so it is all in one place. I would have loved to inline this logic. Unfortunately, statichttprepo also calls it. I didn't want to inline it twice. We could potentially refactor statichttppeer. But meh. Differential Revision: https://phab.mercurial-scm.org/D4574

localrepo: resolve store and cachevfs in makelocalrepository() This is mostly a code move and refactor. One change is that we now explicitly look for requirements indicating a share is being used rather than blindly try to read from .hg/sharedpath. Requirements *should* be all that is necessary to dictate high-level behavior and I'm not sure why the previous code was doing what it was. The previous code has been in place since 87d1fd40f57e (authored in 2009). And the commit immediately after that (971e38a9344b) introduced ``hg.share()`` and always wrote the ``shared`` requirement. And as far as I can tell, every revision of ``hg.share()`` since has written either the ``shared`` or ``relshared`` requirement. So I'm pretty sure we don't need to maintain BC by always looking for and honoring the ``.hg/sharedpath`` file even if a requirement isn't present. .. bc:: A repository will no longer use shared storage if it has a ``.hg/sharedpath`` file but no entry in ``.hg/requires`` saying it is shared. This change should not have any end-user impact, as all shared repos should have a ``.hg/requires`` file indicating this. Differential Revision: https://phab.mercurial-scm.org/D4573

localrepo: document and test bug around opening shared repos As part of refactoring this code, I realized that we don't validate the requirements of a shared repository. This commit documents that next to the requirements validation code and adds a test demonstrating the buggy behavior. I'm not sure if I'll fix this. But it is definitely a bug that users could encounter, as LFS, narrow, and potentially other extensions dynamically add requirements on first use. One part of this I'm not sure about is how to handle loading the .hg/hgrc of the shared repo. We need to do that in order to load extensions. But we don't want that repo's hgrc to overwrite the current repo's. Differential Revision: https://phab.mercurial-scm.org/D4572

localrepo: move requirements reasonability testing to own function Just because we know how to handle each listed requirement doesn't mean that set of requirements is reasonable. This commit introduces an extension-wrappable function to validate that a set of requirements makes sense. We could combine this with ensurerequirementsrecognized(). But I think having a line between basic membership testing and compatibility checking is more powerful as it will help differentiate between missing support and buggy behavior. Differential Revision: https://phab.mercurial-scm.org/D4571

statichttprepo: use new functions for requirements validation The new code in localrepo for requirements gathering and validation is more robust than scmutil.readrequires(). Let's port statichttprepo to it. Since scmutil.readrequires() is no longer used, it has been removed. It is possible extensions were monkeypatching this to supplement the set of supported requirements. But the proper way to do that is to register a featuresetupfuncs. I'm comfortable forcing the API break because featuresetupfuncs is more robust and has been supported for a while. .. api:: ``scmutil.readrequires()`` has been removed. Use ``localrepo.featuresetupfuncs`` to register new repository requirements. Use ``localrepo.ensurerequirementsrecognized()`` to validate them. Differential Revision: https://phab.mercurial-scm.org/D4570

localrepo: validate supported requirements in makelocalrepository() This should be a glorified code move. I did take the opportunity to refactor things. We now have a separate function for gathering requirements and one for validating them. I also mode cosmetic changes to the code, such as not using abbreviations and using a set instead of list to model missing requirements. Differential Revision: https://phab.mercurial-scm.org/D4569

localrepo: read requirements file in makelocalrepository() Previously, scmutil.readrequires() loaded the requirements file and validated its content against what was supported. Requirements translate to repository features and are critical to our plans to dynamically create local repository types. So, we must load them in makelocalrepository() before a repository instance is constructed. This commit moves the reading of the .hg/requires file to makelocalrepository(). Because scmutil.readrequires() was performing I/O and validation, we inlined the validation into localrepository.__init__ and removed scmutil.readrequires(). I plan to remove scmutil.readrequires() in a future commit (we can't do it now because statichttprepo uses it). Differential Revision: https://phab.mercurial-scm.org/D4568

localrepo: check for .hg/ directory in makelocalrepository() As part of this, we move the check to before .hg/hgrc is loaded, as it makes sense to check for the directory before attempting to open a file in it. Differential Revision: https://phab.mercurial-scm.org/D4567

localrepo: load extensions in makelocalrepository() Behavior does change subtly. First, we now load the hgrc before optionally setting up the vfs ward. That's fine: the vfs ward is for debugging and we know we won't hit it when reading .hg/hgrc. If the loaded extension were performing repo/vfs I/O, then we'd be worried. But extensions don't have access to the repo object that loaded them when they are loaded. Unless they are doing stack walking as part of module loading (which would be crazy), they shouldn't have access to the repo that incurred their load. Second, we now load extensions outside of the try..except IOError block. Previously, if loading an extension raised IOError, it would be silently ignored. I'm pretty sure the IOError is there for missing .hgrc files and should never have been ignored for issues loading extensions. I don't think this matters in reality because extension loading traps I/O errors. Differential Revision: https://phab.mercurial-scm.org/D4566

localrepo: copy ui in makelocalrepository() We will want to load the .hg/hgrc file from makelocalrepository() so we can consult its options as part of deriving the repository type. This means we need to create our ui instance copy in that function. Differential Revision: https://phab.mercurial-scm.org/D4565

localrepo: move some vfs initialization out of __init__ In order to make repository types more dynamic, we'll need to move the logic for determining repository behavior out of localrepository.__init__ so we can influence behavior before the type is instantiated. This commit starts that process by moving working directory and .hg/ vfs initialization to our new standalone function for instantiating local repositories. Aside from API changes, behavior should be fully backwards compatible. .. api:: localrepository.__init__ now does less work and accepts new args Use ``hg.repository()``, ``localrepo.instance()``, or ``localrepo.makelocalrepository()`` to obtain a new local repository instance instead of calling the ``localrepository`` constructor directly. Differential Revision: https://phab.mercurial-scm.org/D4564

localrepo: create new function for instantiating a local repo object Today, there is a single local repository class - localrepository. Its __init__ is responsible for loading the .hg/requires file and taking different actions depending on what is present. In addition, extensions may define a "reposetup" function that monkeypatches constructed repository instances, often by implementing a derived type and changing the __class__ of the repo instance. Work around alternate storage backends and partial clone has made it clear to me that shoehorning all this logic into __init__ and operating on an existing instance is too convoluted. For example, localrepository assumes revlog storage and swapping in non-revlog storage requires overriding e.g. file() to return something that isn't a revlog. I've authored various patches that either: a) teach various methods (like file()) about different states and taking the appropriate code path at run-time b) create methods/attributes/callables used for instantiating things and populating these in __init__ "a" incurs run-time performance penalties and makes code more complicated since various functions have a bunch of "if storage is X" branches. "b" makes localrepository quickly explode in complexity. My plan for tackling this problem is to make the local repository type more dynamic. Instead of a static localrepository class/type that supports all of the local repository configurations (revlogs vs other, revlogs with ellipsis, revlog v1 versus revlog v2, etc), we'll dynamically construct a type providing the implementations that are needed for the repository on disk, derived from the .hg/requires file and configuration options. The constructed repository type will be specialized and methods won't need to be taught about different implementations nor overloaded. We may also leverage this functionality for building types that don't implement all attributes. For example, the "intents" feature allows commands to declare that they are read only. By dynamically constructing a repository type, we could return a repository instance with no attributes related to mutating the repository. This could include things like a "changelog" property implementation that doesn't check whether it needs to invalidate the hidden revisions set on every access. This commit establishes a function for building a local repository instance. Future commits will start moving functionality from localrepository.__init__ to this function. Then we'll start dynamically changing the returned type depending on options that are present. This change may seem radical. But it should be fully compatible with the reposetup() model - at least for now. Differential Revision: https://phab.mercurial-scm.org/D4563

transaction: make entries a private attribute (API) This attribute is tracking changes to append-only files. It is an implementation detail and should not be exposed as part of the public interface. But code in repair was accessing it, so it seemingly does belong as part of the public API. But that code in repair is making assumptions about how storage works and is grossly wrong when alternate storage backends are in play. We'll need some kind of "strip" API at the storage layer that knows how to handle things in a storage-agnostic manner. I don't think accessing a private attribute on the transaction is any worse than what this code is already doing. So I'm fine with violating the abstraction for transactions. And with this change, all per-instance attributes on transaction have been made private except for "changes" and "hookargs." Both are used by multiple consumers and look like they need to be part of the public interface. .. api:: Various attributes of ``transaction.transaction`` are now ``_`` prefixed to indicate they shouldn't be used by external consumers. Differential Revision: https://phab.mercurial-scm.org/D4634

transaction: make names a private attribute This is used to report the transaction name in __repr__. It is very obviously an implementation detail and doesn't need to be exposed as part of the public interface. So mark it as private. Differential Revision: https://phab.mercurial-scm.org/D4633

transaction: make map a private attribute This is used to track which files are modified. It is an implementation detail of current transactions and doesn't need to be exposed to the public interface. So mark it as private. Differential Revision: https://phab.mercurial-scm.org/D4632

transaction: make report a private attribute This is a callable used for logging. It isn't used outside the transaction code. It doesn't need to be part of the public interface. Let's mark it as private. Differential Revision: https://phab.mercurial-scm.org/D4631

transaction: make opener a private attribute The VFS instance is an implementation detail of the transaction and doesn't belong as part of the public interface. So mark it as private. Differential Revision: https://phab.mercurial-scm.org/D4630

transaction: make after a private attribute This is another callable that is passed in at __init__ time. It doesn't need to be part of the public interface. Differential Revision: https://phab.mercurial-scm.org/D4629

transaction: make checkambigfiles a private attribute This holds instance state that is passed in at __init__ time. It doesn't need to be exposed as part of the public interface. Differential Revision: https://phab.mercurial-scm.org/D4628

transaction: make validator a private attribute This is similar to releasefn. It holds state that doesn't need to be exposed as part of the public interface. Differential Revision: https://phab.mercurial-scm.org/D4627

transaction: make releasefn a private attribute This is a handle on a callable that is called when the journal is closed. The value is specified at __init__ time. It doesn't need to be exposed on the public interface. So mark it as private. Differential Revision: https://phab.mercurial-scm.org/D4626

transaction: make file a private attribute This holds a file handle for the journal file. This file handle should not be touched outside the journal class and doesn't belong on the public interface. Differential Revision: https://phab.mercurial-scm.org/D4625

transaction: make journal a private attribute This attribute tracks the name of the journal file. It is an implementation detail of the current transaction and therefore shouldn't be exposed as part of the interface. Let's mark it as private. Differential Revision: https://phab.mercurial-scm.org/D4624

transaction: make undoname a private attribute This attribute tracks the file pattern to use for undo files. It is an implementation detail of the current transaction semantics and doesn't need to be part of the future transaction interface. So mark it as private. Differential Revision: https://phab.mercurial-scm.org/D4623

transaction: make count and usages private attributes I want to formalize the interface for transactions. As part of doing that, let's take the opportunity to make some attributes non-public. "count" and "usages" track how many times the transaction has been opened/nested/closed/released. This is internal state and doesn't need to be part of the public API. Differential Revision: https://phab.mercurial-scm.org/D4622

narrow: don't send the changelog information when widening without ellipses When we widen anon-ellipses narrow copy, the server sends the changelog information of all the changesets. The code was copied from ellipses case and in ellipses cases, it's required to send the new changelog data. But in non-ellipses cases, we don't need to send the changelog data as we will have all the changesets locally. Before this patch, there was a overhead of ~8-10 mins on each widening call because of all the changelog information being pulled and being applied. After this patch, we no more pull the changelog information. So this patch can save ~5 mins on Mozilla repo on each widening and more on repos which have more changesets. When we apply an empty changelog from changegroup, there is a devel-warn. This patch kind of hacks to silence that devel-warn. Differential Revision: https://phab.mercurial-scm.org/D4639

changegroup: add functionality to skip adding changelog data to changegroup In narrow extension, when we have a non-ellipses narrow working copy and we extend it, we pull all the changelog data again and the client tries to reapply all that changelog data. While downloading millions of changeset data is still not very expensive but applying them on the client side is very expensive and takes ~10 minutes. These 10 minutes are added to every `hg tracked --addinclude <>` call and extending a narrow copy becomes very slow. This patch adds a new changelog argument to cgpacker.generate() fn. If the changelog argument is set to False, we won't yield the changelog data. We still have to iterate over the deltas returned by _generatechangelog() because that's a generator and builds the data for clstate variable which is required for calculating manifests and filelogs. Differential Revision: https://phab.mercurial-scm.org/D4638

tests: add debug output in test-narrow-widen-no-ellipsis.t This will help us in understanding the upcoming patches better. Differential Revision: https://phab.mercurial-scm.org/D4637

changegroup: improve the devel-warn to specify changelog was empty Right now, the develwarn says "applied empty changegroup" which is not correct because we can send a changegroup without changelog with just manifest and filelogs and it will still say the same. Let's fix this to say that we are applying empty changelog from changegroup. In future patches I am will be adding functionality to send a changegroup from the server without an empty changelog. Differential Revision: https://phab.mercurial-scm.org/D4636

zsh_completion: add -b/--branch and -B/--bookmark(s) flags properly _hg_branch_bmark_opts used to add these two flags, but had the same descriptions for the flags regardless of what command took them and didn't allow specifying flags more than once (no '*' at the start). Even more importantly, it assumed that -B was always expecting an argument (i.e. --bookmark=foo), but in case of incoming and outgoing it's not so (--bookmarks is self-sufficient). Differential Revision: https://phab.mercurial-scm.org/D4612

narrow: when writing treemanifests, skip inspecting directories outside narrow This provides significant speed benefits when narrow and treemanifests are in use, see the timing numbers below. Note that like previously, differences of <5% are considered noise. The below timing numbers are in the same style as previously (example: ee7ee0c516ca). 'before' is 9db85644, and does not include that example commit's improvements. diff --git: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 1.327 s +- 0.051 s | 1.296 s +- 0.009 s | 97.7% m-u | | x | 1.310 s +- 0.020 s | 1.295 s +- 0.015 s | 98.9% m-u | x | | 1.295 s +- 0.018 s | 1.296 s +- 0.007 s | 100.1% m-u | x | x | 83.5 ms +- 0.8 ms | 84.1 ms +- 0.8 ms | 100.7% l-d-r | | | 205.1 ms +- 3.5 ms | 205.0 ms +- 3.8 ms | 100.0% l-d-r | | x | 194.2 ms +- 5.6 ms | 192.3 ms +- 4.3 ms | 99.0% l-d-r | x | | 99.1 ms +- 2.2 ms | 97.8 ms +- 0.9 ms | 98.7% l-d-r | x | x | 66.2 ms +- 1.0 ms | 67.2 ms +- 2.7 ms | 101.5% diff -c . --git: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 233.9 ms +- 1.9 ms | 235.6 ms +- 5.1 ms | 100.7% m-u | | x | 151.4 ms +- 1.2 ms | 152.2 ms +- 2.0 ms | 100.5% m-u | x | | 234.8 ms +- 2.7 ms | 235.0 ms +- 2.7 ms | 100.1% m-u | x | x | 127.8 ms +- 2.1 ms | 126.0 ms +- 1.1 ms | 98.6% l-d-r | | | 82.5 ms +- 1.6 ms | 82.3 ms +- 2.0 ms | 99.8% l-d-r | | x | 3.742 s +- 0.017 s | 3.819 s +- 0.208 s | 102.1% l-d-r | x | | 84.4 ms +- 1.5 ms | 83.2 ms +- 1.0 ms | 98.6% l-d-r | x | x | 751.2 ms +- 5.0 ms | 755.8 ms +- 12.9 ms | 100.6% rebase -r . --keep -d .^^: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 5.519 s +- 0.038 s | 5.526 s +- 0.057 s | 100.1% m-u | | x | 5.588 s +- 0.048 s | 5.607 s +- 0.061 s | 100.3% m-u | x | | 5.520 s +- 0.044 s | 5.546 s +- 0.059 s | 100.5% m-u | x | x | 586.6 ms +- 12.8 ms | 554.9 ms +- 21.2 ms | 94.6% <-- l-d-r | | | 629.8 ms +- 5.5 ms | 627.4 ms +- 6.6 ms | 99.6% l-d-r | | x | 6.165 s +- 0.058 s | 6.255 s +- 0.303 s | 101.5% l-d-r | x | | 270.2 ms +- 2.3 ms | 271.4 ms +- 2.7 ms | 100.4% l-d-r | x | x | 4.700 s +- 0.025 s | 1.651 s +- 0.016 s | 35.1% <-- status --change . --copies: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 215.4 ms +- 2.3 ms | 216.5 ms +- 4.2 ms | 100.5% m-u | | x | 132.9 ms +- 1.2 ms | 132.0 ms +- 1.4 ms | 99.3% m-u | x | | 217.0 ms +- 1.9 ms | 215.4 ms +- 1.9 ms | 99.3% m-u | x | x | 108.6 ms +- 1.0 ms | 108.2 ms +- 1.5 ms | 99.6% l-d-r | | | 80.0 ms +- 1.3 ms | 80.5 ms +- 1.1 ms | 100.6% l-d-r | | x | 3.916 s +- 0.187 s | 3.966 s +- 0.236 s | 101.3% l-d-r | x | | 84.4 ms +- 3.1 ms | 83.9 ms +- 1.1 ms | 99.4% l-d-r | x | x | 758.0 ms +- 8.2 ms | 753.5 ms +- 5.0 ms | 99.4% status --copies: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 1.905 s +- 0.025 s | 1.910 s +- 0.044 s | 100.3% m-u | | x | 1.892 s +- 0.009 s | 1.895 s +- 0.012 s | 100.2% m-u | x | | 1.891 s +- 0.012 s | 1.902 s +- 0.018 s | 100.6% m-u | x | x | 93.3 ms +- 0.9 ms | 93.4 ms +- 0.8 ms | 100.1% l-d-r | | | 570.7 ms +- 7.8 ms | 571.9 ms +- 18.5 ms | 100.2% l-d-r | | x | 561.5 ms +- 5.2 ms | 562.9 ms +- 6.1 ms | 100.2% l-d-r | x | | 171.7 ms +- 2.6 ms | 171.9 ms +- 1.2 ms | 100.1% l-d-r | x | x | 142.7 ms +- 2.0 ms | 140.3 ms +- 1.0 ms | 98.3% update $rev^; ~/src/hg/hg{hg}/hg update $rev: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 3.126 s +- 0.016 s | 3.128 s +- 0.015 s | 100.1% m-u | | x | 3.014 s +- 0.068 s | 3.008 s +- 0.031 s | 99.8% m-u | x | | 3.143 s +- 0.037 s | 3.184 s +- 0.086 s | 101.3% m-u | x | x | 308.0 ms +- 1.8 ms | 308.1 ms +- 5.7 ms | 100.0% l-d-r | | | 430.8 ms +- 4.5 ms | 436.4 ms +- 8.7 ms | 101.3% l-d-r | | x | 9.676 s +- 0.127 s | 9.945 s +- 0.272 s | 102.8% l-d-r | x | | 254.2 ms +- 3.3 ms | 255.7 ms +- 3.1 ms | 100.6% l-d-r | x | x | 1.571 s +- 0.030 s | 1.555 s +- 0.014 s | 99.0% Differential Revision: https://phab.mercurial-scm.org/D4606

tests: fix a couple of drawdag.py references Differential Revision: https://phab.mercurial-scm.org/D4635

py3: fix kwargs handling in hgext/fastannotate.py Differential Revision: https://phab.mercurial-scm.org/D4588

narrow: use diffmatcher to send only new filelogs in non-ellipses widening Before this patch, when we widen a non-ellipses narrow clone, we downloads all the filelogs matching the resulting new matcher. This is same as the ellipses case but can be improved because, we don't pull new csets in non-ellipses cases, we can only download the new added files instead of downloading all the files which matches the new matcher. So, we only download files which matches the new matcher but does not matches the old matcher. There exists a match.differencematcher() which is used here. This will lead to significant amount of speedup in extending a non-ellipses narrow copy on large repos because we will download and process only the new required filelogs. The tests changes demonstrate that we are downloading now less files. Thanks to Augie for pointing that functionality of differencematcher exists in core. Differential Revision: https://phab.mercurial-scm.org/D4614

py3: add missing b'' prefixes in couple of test files These were missed in the earlier patch and caught by Yuya. # skip-blame because just b'' prefix Differential Revision: https://phab.mercurial-scm.org/D4613

run-tests: convert the remaining os.system() call to Unicode I wasn't able to hit this path in 543a788eea2d, but I have now when I accidentally left off `--local`.

py3: partially fix pager spawning on Windows Previously, spinning up the pager crashed because the command and environment was in bytes. (See also 543a788eea2d.) Now it aborts with an invalid handle: $ HGMODULEPOLICY=py py -3 ../hg --traceback --config extensions.evolve=! Traceback (most recent call last): File "c:\Users\Matt\projects\hg\mercurial\ui.py", line 967, in _write self.fout.write(''.join(msgs)) File "c:\Users\Matt\projects\hg\mercurial\windows.py", line 173, in write self.fp.write(s[start:end]) OSError: [WinError 6] The handle is invalid During handling of the above exception, another exception occurred: Traceback (most recent call last): File "c:\Users\Matt\projects\hg\mercurial\scmutil.py", line 164, in callcatch return func() File "c:\Users\Matt\projects\hg\mercurial\dispatch.py", line 350, in _runcatchfunc return _dispatch(req) File "c:\Users\Matt\projects\hg\mercurial\dispatch.py", line 930, in _dispatch return commands.help_(ui, 'shortlist') File "c:\Users\Matt\projects\hg\mercurial\commands.py", line 2930, in help_ ui.write(formatted) File "c:\Users\Matt\projects\hg\mercurial\ui.py", line 948, in write self._writenobuf(*args, **opts) File "c:\Users\Matt\projects\hg\mercurial\ui.py", line 960, in _writenobuf self._write(*msgs, **opts) File "c:\Users\Matt\projects\hg\mercurial\ui.py", line 969, in _write raise error.StdioError(err) mercurial.error.StdioError: [Errno 9] The handle is invalid abort: The handle is invalid The interesting bit here is that the abort message is marked with ANSI color, but the OSError is not.

censor: rename loop variable to silence pyflakes warning hgext/censor.py:92: list comprehension redefines 'c' from line 88

py3: add b'' prefixes in tests/test-hgweb-no-request-uri.t # skip-blame because just b'' prefixes. Differential Revision: https://phab.mercurial-scm.org/D4611

py3: add b'' prefixes in tests/test-hgweb-no-path-info.t # skip-blame because just b'' prefixes Differential Revision: https://phab.mercurial-scm.org/D4610

py3: add b'' prefixes in tests/test-hgweb-non-interactive.t # skip-blame because just b'' prefix Differential Revision: https://phab.mercurial-scm.org/D4609

py3: use codecs.encode() to encode in rot-13 encoding The other occurence will need some more love as description is bytes by default and we need to decode it and then encode it. Differential Revision: https://phab.mercurial-scm.org/D4608

py3: add two passing tests to whitelist found by buildbot The buildbot found these two new passing tests on Python 3. Differential Revision: https://phab.mercurial-scm.org/D4607

phabricator: mark extension as experimental for now I don't want us to commit to this having a stable interface just yet. Differential Revision: https://phab.mercurial-scm.org/D4605

phabricator: fix templating bug by using hybriddict Differential Revision: https://phab.mercurial-scm.org/D4604

phabricator: add tests of templatekeyword Having tests is paying off: I found a bug and now it'll be easy to fix! Differential Revision: https://phab.mercurial-scm.org/D4603

phabricator: move extension from contrib to hgext It's well-enough tested now and widely enough used I think we should ship it. Differential Revision: https://phab.mercurial-scm.org/D4602

tests: add some basic tests of phabricator interactions This uses the vcr library to avoid hitting phabricator on every test execution. In order to generate new recordings (vcr calls them cassettes) just remove the appropriate json file, and the test will regenerate it. It's not my favorite way to test things, but it'll let us have test coverage on the phabricator extension that'll make it resilient to refactors in core and let us move it to hgext. In the future, it'd probably be better to have a docker container we can spin up for creating the vcr recordings, but for now this is enough better than nothing I'm going to declare victory. Coverage reports about 73% of the extension is now covered. Differential Revision: https://phab.mercurial-scm.org/D4601

phabricator: add support for using the vcr library to mock interactions I'll use this in an upcoming test. The decorator dancing in this is more complicated than I'd like, but it beats repeating all this code everywhere. Differential Revision: https://phab.mercurial-scm.org/D4600

keepalive: work around slight deficiency in vcr VCR's response type doesn't define the will_close attribute. Let's just have keepalive default to closing the socket if the will_close attribute is missing. Differential Revision: https://phab.mercurial-scm.org/D4599

hghave: add a checker for the vcr HTTP record/replay library I'm going to use this to write some tests of the phabricator extension. Differential Revision: https://phab.mercurial-scm.org/D4598

py3: allow run-tests.py to run on Windows This is now functional: HGMODULEPOLICY=py py -3 run-tests.py --local test-help.t --pure --view bcompare However, on this machine without a C compiler, it tries to load cext anyway, and blows up. I haven't looked into why, other than to see that it does set the environment variable. When the test exits though, I see it can't find killdaemons.py, get-with-headers.py, etc. I have no idea why these changes are needed, given that it runs on Linux. But os.system() is insisting that it take a str, and subprocess.Popen() blows up without str: Errored test-help.t: Traceback (most recent call last): File "run-tests.py", line 810, in run self.runTest() File "run-tests.py", line 858, in runTest ret, out = self._run(env) File "run-tests.py", line 1268, in _run exitcode, output = self._runcommand(cmd, env) File "run-tests.py", line 1141, in _runcommand env=env) File "C:\Program Files\Python37\lib\subprocess.py", line 756, in __init__ restore_signals, start_new_session) File "C:\Program Files\Python37\lib\subprocess.py", line 1100, in _execute_child args = list2cmdline(args) File "C:\Program Files\Python37\lib\subprocess.py", line 511, in list2cmdline needquote = (" " in arg) or ("\t" in arg) or not arg TypeError: argument of type 'int' is not iterable This is exactly how it crashes when trying to spin up a pager too. I left one instance of os.system() unchanged in _installhg(), because it doesn't get there.

py3: ensure run-tests environment is uniformly str subprocess.popen() was crashing, and when I printed out `env`, all of the keys and most of the values were str. Except these.

py3: ensure run-tests.osenvironb is actually bytes Windows doesn't have os.environb, so it was falling back to the Unicode form, and all of the accesses are trying to use bytes.

py3: fix str vs bytes in enough places to run `hg version` on Windows I don't have Visual Studio 2015 at home, but this now works with a handful of extensions (blackbox, extdiff, patchbomb, phabricator and rebase, but not evolve): $ HGMODULEPOLICY=py py -3 ../hg version Enabling the evolve extension causes the usual "failed to import ..." line, but then print this before the usual version output: ('commit', '[b'debugancestor', b'debugapplystreamclonebundle', ..., b'verify', b'version']') ... where the elided part seems to be every command and alias known.

windows: open registry keys using unicode names Python3 complained it must be str. While here, use a context manager to close the key- it wouldn't wrap at 80 characters the old way, and would have had to move anyway.

py3: byteify strings in pycompat These surfaced when disabling the source transformer to debug the problems in win32.py. ./contrib/byteify-strings.py found a couple false positives, so I marked them with r'' explicitly (in case I'm wrong). # skip-blame since this is just b'' and r'' prefixing

wireprotov2: let clients drive delta behavior Previously, the "manifestdata" and "filedata" commands assumed the receiver had all parent revisions for requested nodes. Unless the revision had no parents, they emitted a delta instead of a fulltext. This strategy isn't appropriate for shallow clones and for clients that only want to access fulltext revision data for a single node without fetching their parent revisions. This commit adds an "haveparents" argument to the "manifestdata" and "filedata" commands that controls delta generation behavior. Unless "haveparents" is set, the server assumes that the client doesn't have parent revisions unless they were previously sent as part of the current group of revisions. This change allows the fulltext revision data of any individual revision to be obtained. This will facilitate shallow clones and other data retrieval strategies that don't require all previous revisions of an entity to be fetched. Differential Revision: https://phab.mercurial-scm.org/D4492

exchangev2: fetch file revisions Now that the server has an API for fetching file data, we can call into it to fetch file revisions. The implementation is relatively straightforward: we examine the manifests that we fetched and find all new file revisions referenced by them. We build up a mapping from file path to file nodes to manifest node. (The mapping to first manifest node allows us to map back to first changelog node/revision, which is used for the linkrev.) Once that map is built up, we iterate over it in a deterministic manner and fetch and store file data. The code is very similar to manifest fetching. So similar that we could probably extract the common bits into a generic function. With file data retrieval implemented, `hg clone` and `hg pull` are effectively feature complete, at least as far as the completeness of data transfer for essential repository data (changesets, manifests, files, phases, and bookmarks). We're still missing support for obsolescence markers, the hgtags fnodes cache, and the branchmap cache. But these are non-essential for the moment (and will be implemented later). This is a good point to assess the state of exchangev2 in terms of performance. I ran a local `hg clone` for the mozilla-unified repository using both version 1 and version 2 of the wire protocols and exchange methods. This is effectively comparing the performance of the wire protocol overhead and "getbundle" versus domain-specific commands. Wire protocol version 2 doesn't have compression implemented yet. So I tested version 1 with `server.compressionengines=none` to remove compression overhead from the equation. server before: user 220.420+0.000 sys 14.420+0.000 after: user 321.980+0.000 sys 18.990+0.000 client before: real 561.650 secs (user 497.670+0.000 sys 28.160+0.000) after: real 1226.260 secs (user 944.240+0.000 sys 354.150+0.000) We have substantial regressions on both client and server. This is obviously not desirable. I'm aware of some reasons: * Lack of hgtagsfnodes transfer (contributes significant CPU to client). * Lack of branch cache transfer (contributes significant CPU to client). * Little to no profiling / optimization performed on wire protocol version 2 code. * There appears to be a memory leak on the client and that is likely causing swapping on my machine. * Using multiple threads on the client may be counter-productive because Python. * We're not compressing on the server. * We're tracking file nodes on the client via manifest diffing rather than using linkrev shortcuts on the server. I'm pretty confident that most of these issues are addressable. But even if we can't get wire protocol version 2 on performance parity with "getbundle," I still think it is important to have the set of low level data-specific retrieval commands that we have implemented so far. This is because the existence of such commands allows flexibility in how clients access server data. Differential Revision: https://phab.mercurial-scm.org/D4491

wireprotov2: define and implement "filedata" command Continuing our trend of implementing *data commands for retrieving information about specific repository data primitives, this commit implements a command for retrieving data about an individual tracked file. The command is very similar to "manifestdata." The only significant difference is that we have a standalone function for obtaining storage for a tracked file. This is to provide a monkeypatch point for extensions to implement path-based access control. With this API available, wire protocol version 2 now exposes all data primitives necessary to implement a full clone. Of course, since "filedata" can only resolve data for a single path at a time, clients would need to issue N commands to perform a full clone. On the Firefox repository, this would be ~461k commands. We'll likely need to implement a file data retrieval command that supports multiple paths. But that can be implemented later. Differential Revision: https://phab.mercurial-scm.org/D4490

exchangev2: fetch manifest revisions Now that the server has support for retrieving manifest data, we can implement the client bits to call it. We teach the changeset fetching code to capture the manifest revisions that are encountered on incoming changesets. We then feed this into a new function which filters out known manifests and then batches up manifest data requests to the server. This is different from the previous wire protocol in a few notable ways. First, the client fetches manifest data separately and explicitly. Before, we'd ask the server for data pertaining to some changesets (via a "getbundle" command) and manifests (and files) would be sent automatically. Providing an API for looking up just manifest data separately gives clients much more flexibility for manifest management. For example, a client may choose to only fetch manifest data on demand instead of prefetching it (i.e. partial clone). Second, we send N commands to the server for manifest retrieval instead of 1. This property has a few nice side-effects. One is that the deterministic nature of the requests lends itself to server-side caching. For example, say the remote has 50,000 manifests. If the server is configured to cache responses, each time a new commit arrives, you will have a cache miss and need to regenerate all outgoing data. But if you makes N requests requesting 10,000 manifests each, a new commit will still yield cache hits on the initial, unchanged manifest batches/requests. A derived benefit from these properties is that resumable clone is conceptually simpler to implement. When making a monolithic request for all of the repository data, recovering from an interrupted clone is hard because the server was in the driver's seat and was maintaining state about all the data that needed transferred. With the client driving fetching, the client can persist the set of unfetched entities and retry/resume a fetch if something goes wrong. Or we can fetch all data N changesets at a time and slowly build up a repository. This approach is drastically easier to implement when we have server APIs exposing low-level repository primitives (such as manifests and files). We don't yet support tree manifests. But it should be possible to implement that with the existing wire protocol command. Differential Revision: https://phab.mercurial-scm.org/D4489

wireprotov2: define and implement "manifestdata" command The added command can be used for obtaining manifest data. Given a manifest path and set of manifest nodes, data about manifests can be retrieved. Unlike changeset data, we wish to emit deltas to describe manifest revisions. So the command uses the relatively new API for building delta requests and emitting them. The code calls into deltaparent(), which I'm not very keen of. There's still work to be done in delta generation land so implementation details of storage (e.g. exactly one delta is stored/available) don't creep into higher levels. But we can worry about this later (there is already a TODO on imanifestorage tracking this). On the subject of parent deltas, the server assumes parent revisions exist on the receiving end. This is obviously wrong for shallow clone. I've added TODOs to add a mechanism to the command to allow clients to specify desired behavior. This shouldn't be too difficult to implement. Another big change is that the client must explicitly request manifest nodes to retrieve. This is a major departure from "getbundle," where the server derives relevant manifests as it iterates changesets and sends them automatically. As implemented, the client must transmit each requested node to the server. At 20 bytes per node, we're looking at 2 MB per 100,000 nodes. Plus wire encoding overhead. This isn't ideal for clients with limited upload bandwidth. I plan to address this in the future by allowing alternate mechanisms for defining the revisions to retrieve. One idea is to define a range of changeset revisions whose manifest revisions to retrieve (similar to how "changesetdata" works). We almost certainly want an API to look up an individual manifest by node. And that's where I've chosen to start with the implementation. Again, a theme of this early exchangev2 work is I want to start by building primitives for accessing raw repository data first and see how far we can get with those before we need more complexity. Differential Revision: https://phab.mercurial-scm.org/D4488

wireprotov2: add TODOs around extending changesetdata fields Extensions will inevitably want to extend the set of changeset data/fields that can be requested. We'll need to implement support for extending this in the future. Add some TODOs to track that. Differential Revision: https://phab.mercurial-scm.org/D4487

exchangev2: fetch and apply bookmarks This is pretty similar to phases data. We collect bookmarks data as we process records. Then at the end we make a call to the bookmarks subsystem to reflect the remote's bookmarks. Like phases, the code for handling bookmarks is vastly simpler than the previous wire protocol code because the server always transfers the full set of bookmarks when bookmarks are requested. We don't have to keep track of whether we requested bookmarks or not. Differential Revision: https://phab.mercurial-scm.org/D4486

wireprotov2: add bookmarks to "changesetdata" command Like we did for phases, we want to emit bookmarks data attached to each changeset. The approach here is very similar to phases: we emit bookmarks data inline with requested revision data. But we emit records for nodes that weren't requested as well so consumers have access to the full set of defined bookmarks. Differential Revision: https://phab.mercurial-scm.org/D4485

exchangev2: fetch and apply phases data Now that the server supports emitting phases data, we can request it and apply it on the client. Because we may receive phases-only updates from the server, we no longer conditionally perform the "changesetdata" command depending on whether there are revisions to fetch. In the previous wire protocol, this case would result in us falling back to performing "listkeys" commands to look up phases, bookmarks, etc data. But since "changesetdata" is smart enough to handle metadata only fetches, we can keep things consistent. It's worth noting that because of the unified approach to changeset data retrieval, phase handling code in wire proto v2 exchange is drastically simpler. Contrast with all the code in exchange.py dealing with all the variations for obtaining phases data. Differential Revision: https://phab.mercurial-scm.org/D4484

wireprotov2: add phases to "changesetdata" command This commit teaches the "changesetdata" wire protocol command to emit the phase state for each changeset. This is a different approach from existing phase transfer in a few ways. Previously, if there are no new revisions (or we're not using bundle2), we perform a "listkeys" request to retrieve phase heads. And when revision data is being transferred with bundle2, phases data is encoded in a standalone bundle2 part. In both cases, phases data is logically decoupled from the changeset data and is encountered/applied after changeset revision data is received. The new wire protocol purposefully tries to more tightly associate changeset metadata (phases, bookmarks, obsolescence markers, etc) with the changeset revision and index data itself, rather than have it live as a separate entity that must be fetched and processed separately. I reckon that one reason we didn't do this before was it was difficult to add new data types/fields without breaking existing consumers. By using CBOR maps to transfer changeset data and putting clients in control of what fields are requested / present in those maps, we can easily add additional changeset data while maintaining backwards compatibility. I believe this to be a superior approach to the problem. That being said, for performance reasons, we may need to resort to alternative mechanisms for transferring data like phases. But for now, I think giving the wire protocol the ability to transfer changeset metadata next to the changeset itself is a powerful feature because it is a raw, changeset-centric data API. And if you build simple APIs for accessing the fundamental units of repository data, you enable client-side experimentation (partial clone, etc). If it turns out that we need specialized APIs or mechanisms for transferring data like phases, we can build in those APIs later. For now, I'd like to see how far we can get on simple APIs. It's worth noting that when phase data is being requested, the server will also emit changeset records for nodes in the bases specified by the "noderange" argument. This is to ensure that phase-only updates for nodes the client has are available to the client, even if no new changesets will be transferred. Differential Revision: https://phab.mercurial-scm.org/D4483

exchangev2: fetch changeset revisions All Mercurial repository data is derived from changesets: you can't do anything unless you have changesets. Therefore, it makes sense for changesets to be the first piece of data that we transfer as part of pull. To do this, we call our new "changesetdata" command, requesting parents and revision data. This gives us all the data that a changegroup delta group would give us. We simply normalize this data into what addgroup() expects and call that API on the changelog to bulk insert revisions into the changelog. Code in this commit is heavily borrowed from changegroup.cg1unpacker.apply(). Differential Revision: https://phab.mercurial-scm.org/D4482

wireprotov2: define and implement "changesetdata" command This commit introduces the "changesetdata" wire protocol command. The role of the command is to expose data associated with changelog revisions, including the raw revision data itself. This command is the first piece of a new clone/pull strategy that is built on top of domain-specific commands for data retrieval. Instead of a monolithic "getbundle" command that transfers all of the things, we'll be introducing commands for fetching specific pieces of data. Since the changeset is the fundamental unit from which we derive pointers to other data (manifests, file nodes, etc), it makes sense to start reimplementing pull with this data. The command accepts as arguments a set of root and head revisions defining the changesets that should be fetched as well as an explicit list of nodes. By default, the command returns only the node values: the client must explicitly request additional fields be added to the response. Current supported fields are the list of parent nodes and the revision fulltext. My plan is to eventually add support for transferring other data associated with changesets, including phases, bookmarks, obsolescence markers, etc. Since the response format is CBOR, we'll be able to add this data into the response object relatively easily (it should be as simple as adding a key in a map). The documentation captures a number of TODO items. Some of these may require BC breaking changes. That's fine: wire protocol v2 is still highly experimental. Differential Revision: https://phab.mercurial-scm.org/D4481

exchangev2: start to implement pull with wire protocol v2 Wire protocol version 2 will take a substantially different approach to exchange than version 1 (at least as far as pulling is concerned). This commit establishes a new exchangev2 module for holding code related to exchange using wire protocol v2. I could have added things to the existing exchange module. But it is already quite big. And doing things inline isn't in question because the existing code is already littered with conditional code for various states of support for the existing wire protocol as it evolved over 10+ years. A new module gives us a chance to make a clean break. This approach does mean we'll end up writing some duplicate code. And there's a significant chance we'll miss functionality as code is ported. The plan is to eventually add #testcase's to existing tests so the new wire protocol is tested side-by-side with the existing one. This will hopefully tease out any features that weren't ported properly. But before we get there, we need to build up support for the new exchange methods. Our journey towards implementing a new exchange begins with pulling. And pulling begins with discovery. The discovery code added to exchangev2 is heavily drawn from the following functions: * exchange._pulldiscoverychangegroup * discovery.findcommonincoming For now, we build on top of existing discovery mechanisms. The new wire protocol should be capable of doing things more efficiently. But I'd rather defer on this problem. To foster the transition, we invent a fake capability on the HTTPv2 peer and have the main pull code in exchange.py call into exchangev2 when the new wire protocol is being used. Differential Revision: https://phab.mercurial-scm.org/D4480

httppeer: expose capabilities for each command This will help code using peers to sniff out exactly what servers support. Differential Revision: https://phab.mercurial-scm.org/D4436

narrow: intersect provided matcher with narrowmatcher in `hg diff` This provides significant speedups when running diff, and no change in behavior that I'm aware of (or that the tests found). I tested with a repo that I started using narrow in after it was created and attempted to run `hg diff -c .` and similar commands in it on a commit that had files not in the narrowspec. Timing numbers below, using a similar setup as my previous commits. before=9db85644, m-u is mozilla-unified at eb39298e432d (flatmanifest) and 0553b7f29eaf (treemanifest). l-d-r is a repo simulating a situation I've encountered where there's one directory with 30k+ subdirectories. N means narrow, T means treemanifest. The narrowspec is pretty small when in use, and importantly the narrowspec is applied *after* doing the initial checkout (without narrowing), so all of these files exist in the filesystem, which is not normally the case if someone has been using narrow for the entire life of the clone. Anything less than a 5% difference in performance is most likely noise. diff --git: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 1.292 s +- 0.009 s | 1.295 s +- 0.010 s | 100.2% m-u | | x | 1.296 s +- 0.042 s | 1.299 s +- 0.026 s | 100.2% m-u | x | | 1.292 s +- 0.010 s | 1.297 s +- 0.021 s | 100.4% m-u | x | x | 84.2 ms +- 1.2 ms | 83.6 ms +- 0.2 ms | 99.3% l-d-r | | | 188.7 ms +- 2.7 ms | 188.8 ms +- 2.0 ms | 100.1% l-d-r | | x | 189.9 ms +- 1.5 ms | 189.4 ms +- 1.2 ms | 99.7% l-d-r | x | | 97.1 ms +- 1.0 ms | 87.1 ms +- 1.0 ms | 89.7% <-- l-d-r | x | x | 96.9 ms +- 0.8 ms | 87.2 ms +- 0.7 ms | 90.0% <-- diff -c . --git: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 231.6 ms +- 3.1 ms | 228.9 ms +- 1.6 ms | 98.8% m-u | | x | 150.5 ms +- 1.7 ms | 150.7 ms +- 1.4 ms | 100.1% m-u | x | | 233.7 ms +- 2.4 ms | 232.2 ms +- 1.9 ms | 99.4% m-u | x | x | 126.1 ms +- 1.2 ms | 126.8 ms +- 1.2 ms | 100.6% l-d-r | | | 82.1 ms +- 2.0 ms | 81.8 ms +- 1.4 ms | 99.6% l-d-r | | x | 3.732 s +- 0.020 s | 3.746 s +- 0.027 s | 100.4% l-d-r | x | | 83.1 ms +- 0.8 ms | 107.6 ms +- 2.4 ms | 129.5% <-- l-d-r | x | x | 758.2 ms +- 38.8 ms | 188.5 ms +- 1.8 ms | 24.9% <-- rebase -r . --keep -d .^^: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 5.532 s +- 0.087 s | 5.496 s +- 0.016 s | 99.3% m-u | | x | 5.554 s +- 0.061 s | 5.532 s +- 0.013 s | 99.6% m-u | x | | 5.602 s +- 0.134 s | 5.508 s +- 0.035 s | 98.3% m-u | x | x | 582.2 ms +- 15.2 ms | 572.9 ms +- 12.0 ms | 98.4% l-d-r | | | 629.5 ms +- 12.3 ms | 622.5 ms +- 7.3 ms | 98.9% l-d-r | | x | 6.173 s +- 0.062 s | 6.185 s +- 0.076 s | 100.2% l-d-r | x | | 274.5 ms +- 10.0 ms | 272.1 ms +- 6.2 ms | 99.1% l-d-r | x | x | 4.835 s +- 0.056 s | 4.826 s +- 0.034 s | 99.8% status --change . --copies: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 214.4 ms +- 1.4 ms | 212.2 ms +- 1.7 ms | 99.0% m-u | | x | 130.9 ms +- 1.2 ms | 131.7 ms +- 1.1 ms | 100.6% m-u | x | | 215.0 ms +- 2.1 ms | 214.9 ms +- 2.7 ms | 100.0% m-u | x | x | 109.5 ms +- 2.3 ms | 107.8 ms +- 0.9 ms | 98.4% l-d-r | | | 79.6 ms +- 0.9 ms | 79.8 ms +- 1.6 ms | 100.3% l-d-r | | x | 3.799 s +- 0.037 s | 3.928 s +- 0.021 s | 103.4% <--? l-d-r | x | | 82.7 ms +- 0.7 ms | 83.2 ms +- 1.0 ms | 100.6% l-d-r | x | x | 746.8 ms +- 6.1 ms | 739.0 ms +- 4.2 ms | 99.0% status --copies: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 1.884 s +- 0.012 s | 1.885 s +- 0.013 s | 100.1% m-u | | x | 1.897 s +- 0.027 s | 1.909 s +- 0.077 s | 100.6% m-u | x | | 1.886 s +- 0.021 s | 1.891 s +- 0.030 s | 100.3% m-u | x | x | 92.0 ms +- 0.7 ms | 92.4 ms +- 0.4 ms | 100.4% l-d-r | | | 570.3 ms +- 18.7 ms | 552.2 ms +- 4.5 ms | 96.8% l-d-r | | x | 568.9 ms +- 16.1 ms | 567.2 ms +- 11.9 ms | 99.7% l-d-r | x | | 171.1 ms +- 2.5 ms | 170.4 ms +- 1.2 ms | 99.6% l-d-r | x | x | 171.6 ms +- 3.4 ms | 171.5 ms +- 1.7 ms | 99.9% update $rev^; ~/src/hg/hg{hg}/hg update $rev: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 3.107 s +- 0.017 s | 3.116 s +- 0.012 s | 100.3% m-u | | x | 2.943 s +- 0.010 s | 2.945 s +- 0.019 s | 100.1% m-u | x | | 3.116 s +- 0.033 s | 3.118 s +- 0.027 s | 100.1% m-u | x | x | 318.5 ms +- 2.7 ms | 320.8 ms +- 4.8 ms | 100.7% l-d-r | | | 428.9 ms +- 4.4 ms | 429.5 ms +- 4.0 ms | 100.1% l-d-r | | x | 9.593 s +- 0.081 s | 9.869 s +- 0.043 s | 102.9% l-d-r | x | | 253.2 ms +- 3.6 ms | 254.0 ms +- 2.8 ms | 100.3% l-d-r | x | x | 1.613 s +- 0.009 s | 1.630 s +- 0.017 s | 101.1% Differential Revision: https://phab.mercurial-scm.org/D4587

identify: change {parents} to a list of nodes (BC) This is a part of the name unification. {parents} is a list of nodes in "hg log -Tjson" output. Since {rev} can be computed from (repo, node) pair, we no longer need to put it to provide {rev} to user templates. https://www.mercurial-scm.org/wiki/GenericTemplatingPlan#Dictionary

identify: use fm.hexfunc thoroughly This fixes the length of {id} in JSON and template outputs.

formatter: replace contexthint() with demand loading of ctx object And pass in repo instead to resolve ctx from (repo, node) pair.

formatter: populate ctx from repo and node value This will basically replace the fm.contexthint() API. I originally thought this would be too complicated, and I wrote 8399438bc7ef "formatter: provide hint of context keys required by template" because of that. However, I had to add a similar mechanism for fctx templates, and the overall machinery became way simpler than my original patch. The test output slightly changed as {author} is no longer available in the {manifest} context, which isn't the point this test targeted on.

merge with stable

py3: call hgweb.hgweb() with bytes values # skip-blame because just b'' prefixes I believe this should fix some tests. Differential Revision: https://phab.mercurial-scm.org/D4594

py3: use '%d' for integers instead of '%s' Differential Revision: https://phab.mercurial-scm.org/D4593

py3: use "%f" for floats instead of "%s" Differential Revision: https://phab.mercurial-scm.org/D4592

py3: suppress the return value from .write() call Differential Revision: https://phab.mercurial-scm.org/D4591

py3: add b'' prefixes in tests/test-diff-color.t # skip-blame because just b'' prefixes Differential Revision: https://phab.mercurial-scm.org/D4590

py3: slice through bytes to prevent getting ascii value I still don't know why python-dev thought it was a nice idea to do this. Differential Revision: https://phab.mercurial-scm.org/D4589

censor: use a reasonable amount of memory Before this change, trying to censor some random revision uses an ever increasing amount of memory (I stopped at 20GB, but it was by no means finished), presumably because these contexts have a lot of information that is kept alive. After this change, the memory usage plateaus quickly. Differential Revision: https://phab.mercurial-scm.org/D4582

help: add internals.wireprotocolrpc to the table

setup: exclude vendored futures package on Python 3 The vendored future can't live on Python 3.

py3: whitelist another passing test Differential Revision: https://phab.mercurial-scm.org/D4562

py3: prevent the win32 ctype _fields_ from being transformed to bytes Otherwise, any hg invocation dies with TypeError: '_fields_' must be a sequence of (name, C type) pairs # skip-blame just a r prefix

cext: fix warnings when building for py3 on Windows MSVC++ 14 now has standard int types that don't need to be redefined (I didn't go back to see when they came along since the build system wants either 2008 or 2015), but doesn't have ssize_t. The FILE pointer in posixfile is only used on python2.

cext: stop preprocessing a partial function call MSVC++ 14 yelled: mercurial/cext/revlog.c(1913): fatal error C1057: unexpected end of file in macro expansion At this point, the C extensions build (with warnings), and it dies in win32.py because the `_fields_` strings in the ctypes classes are being converted to bytes by the source translator.

py3: add b'' to some setup.py strings for Windows These were things found trying to do `make PYTHON="py -3" local`. The following is dumped out, before dying while compiling the C extensions: C:\Program Files\Python37\lib\site-packages\setuptools\dist.py:406: UserWarning: The version specified (b'4.7.1') is an invalid version, this may not work as expected with newer versions of setuptools, pip, and PyPI. Please see PEP 440 for more details. "details." % self.metadata.version running build_py byte-compiling .\mercurial\thirdparty\concurrent\futures\_base.py to _base.cpython-37.pyc File "mercurial\thirdparty\concurrent\futures\_base.py", line 416 raise exception_type, self._exception, self._traceback ^ SyntaxError: invalid syntax # skip-blame since these are just converting to bytes literals

dagop: fix typo spotted while doing unrelated investigation Differential Revision: https://phab.mercurial-scm.org/D4584

hg: don't reuse repo instance after unshare() Unsharing a repository is a pretty invasive procedure and fundamentally changes the behavior of the repository. Currently, hg.unshare() calls into localrepository.__init__ to re-initialize the repository instance. This is a bit hacky. And future commits that refactor how localrepository instances are constructed will make this difficult to support. This commit changes unshare() so it constructs a new repo instance once the unshare I/O has completed. It then poisons the old repo instance so any further use will result in error. Surprisingly, nothing in core appears to access a repo instance after it has been unshared! .. api:: ``hg.unshare()`` now poisons the repo instance so it can't be used. It also returns a new repo instance suitable for interacting with the unshared repository. Differential Revision: https://phab.mercurial-scm.org/D4557

unionrepo: dynamically create repository type from base repository This is basically the same thing we just did for bundlerepo except for union repositories. .. api:: ``unionrepo.unionrepository()`` is no longer usable on its own. To instantiate an instance, call ``unionrepo.instance()`` or ``unionrepo.makeunionrepository()``. Differential Revision: https://phab.mercurial-scm.org/D4556

bundlerepo: dynamically create repository type from base repository Previously, bundlerepository inherited from localrepo.localrepository. You simply instantiated a bundlerepository and its __init__ called localrepo.localrepository.__init__. Things were simple. Unfortunately, this strategy is limiting because it assumes that the base repository is a localrepository instance. And it assumes various properties of localrepository, such as the arguments its __init__ takes. And it prevents us from changing behavior of localrepository.__init__ without also having to change derived classes. Previous and ongoing work to abstract storage revealed these limitations. This commit changes the initialization strategy of bundle repositories to dynamically create a type to represent the repository. Instead of a static type, we instantiate a new local repo instance via localrepo.instance(). We then combine its __class__ with bundlerepository to produce a new type. This ensures that no matter how localrepo.instance() decides to create a repository object, we can derive a bundle repo object from it. i.e. localrepo.instance() could return a type that isn't a localrepository and it would "just work." Well, it would "just work" if bundlerepository's custom implementations only accessed attributes in the documented repository interface. I'm pretty sure it violates the interface contract in a handful of places. But we can worry about that another day. This change gets us closer to doing more clever things around instantiating repository instances without having to worry about teaching bundlerepository about them. .. api:: ``bundlerepo.bundlerepository`` is no longer usable on its own. The class is combined with the class of the base repository it is associated with at run-time. New bundlerepository instances can be obtained by calling ``bundlerepo.instance()`` or ``bundlerepo.makebundlerepository()``. Differential Revision: https://phab.mercurial-scm.org/D4555

bundlerepo: factor out code for instantiating a bundle repository This code will soon become a bit more complicated. So extract to its own function. And change both instantiators of bundlerepository to use it. Differential Revision: https://phab.mercurial-scm.org/D4554

bundlerepo: pass create=True I don't want to know how this came to be. Maybe a holdover from the days before Python had a bool type? Differential Revision: https://phab.mercurial-scm.org/D4553

shelve: use bundlerepo.instance() to construct a repo object The instance() functions are preferred over cls.__init__ for creating repo instances. It doesn't really matter now. But future commits will refactor the bundlerepository class in ways that will cause the old way to break. Differential Revision: https://phab.mercurial-scm.org/D4552

templatekw: add experimental {status} keyword This is another example of fctx-based keywords. I think this is somewhat useful in log templates.

templatekw: add option to include ignored/clean/unknown files in cache They will be necessary to provide {status} of files.

templatekw: keep status tuple in cache dict and rename cache key accordingly There's no point to drop tail elements, which are mostly empty lists.

templatekw: extract function that computes and caches file status

py3: use sysstr() to convert ProgrammingError bytes with no unicode error risk msg.decode('utf8') may fail if msg isn't an ASCII string, and that's possible as we sometimes embed a filename in the error message for example.

revlog: reuse cached delta for identical base revision (issue5975) Since 8f83a953dddf, we skip over empty deltas when choosing a delta base. Such delta happens when two distinct revisions have the same content. The remote might be sending a delta against such revision within the bundle. In that case, the delta base is no longer considered, but the cached one could still, be used with the equivalent revision. Not reusing the delta from the bundle can have a significant performance impact, so we now make sure with doing so when possible.

snapshot: fix line order when skipping over empty deltas The code movement in 37957e07138c introduced an error. Since 8f83a953dddf, we discarded some revisions because they are identical to their delta base (and use that delta base instead). That logic is good, however, in 37957e07138c we mixed up the order of two line, adding the "new" revision to the set of already tested one, instead of the discarded one. So in practice, we were never investigating any revisions in a chain starting with an empty delta. Creating significantly worst delta chain (eg: Mercurial's manifest move goes from about 60MB up to about 80MB).

tests: stabilize change for handling not quoting non-empty-directory The change originated in cb1329738d64. I suspect the problem is with the combination of (re) and the '\' to '/' retry on Windows. I've no idea if py3 on Windows needs the quoting, since it can't even run `hg` with no arguments. (It's dying somewhere on the ctype declarations when win32.py is imported.)

hg: wrap the highest layer in the `hg` script possible in trace event This should help us have a better idea of what "interpreter startup costs" look like. This does omit the HGUNICODEPEDANTRY block and the LIBDIR dancing to set up sys.path, but the former is usually off and the latter is unavoidable and should be very fast. If we get worried about those cases we can consider open-coding the tracing logic here. Differential Revision: https://phab.mercurial-scm.org/D4346

localrepo: use urllocalpath() for path to create repo too It looks like this was lost in 7ce9dea3a14a (localrepo: move repo creation logic out of localrepository.__init__ (API), 2018-09-11). I don't know when it makes a difference (maybe on Windows, since urllocalpath() mentions something about drive letters). Differential Revision: https://phab.mercurial-scm.org/D4550

localrepo: move check for existing repo into createrepository() For symmetry with the check for existence of a repo in localrepository.__init__, we should check for the non-existence in createrepository(). We could alternatively move both checks into instance(). Differential Revision: https://phab.mercurial-scm.org/D4549

py3: add b'' to some run-tests.py strings for Windows Things go seriously off the rails after this, so there may be more that are missing. # skip-blame since these are just converting to bytes literals

wireprotov1peer: forward __name__ of wrapped method in batchable decorator Not required, but clarifies debugging when the going gets really tough. Differential Revision: https://phab.mercurial-scm.org/D4551

templatekw: add {size} keyword as an example of fctx-based keyword I'll add {status}, and I think some lfs keywords can be migrated to this. I'm not certain how many fctx-based keywords will be introduced into the global space, but if there are a couple more, we'll probably need to sort them out to the "File Keywords" section in the templater help. Until then, fctx keywords are hidden as experimental.

formatter: populate fctx from ctx and path value Tests will be added by the next patch.

formatter: factor out function that detects node change and document it This prepares for demand loading of ctx/fctx objects. With this change, 'revcache' is also recreated if 'node' value changes, which will be needed to support loading of ctx from (repo, node) pair.

formatter: inline _gettermap and _knownkeys

formatter: fill missing resources by formatter, not by resource mapper While working on demand loading of ctx/fctx objects, I found it's weird to support lookup in both directions. For instance, fctx can be loaded from (ctx, path) pair, but ctx may also be derived from fctx.changectx() in the original mapping. If the original mapping has had fctx but no ctx, and if the new mapping provides {path}, we can't be sure if fctx should be updated by fctx'.changectx()[path] or not. This patch simply drops the support for the resolution in fctx -> ctx -> repo direction.

templater: remove unused context argument from most resourcemapper functions While working on demand loading of ctx/fctx objects, I noticed that it's quite easy to create infinite recursion by carelessly using the template context in the resource mapper. Let's make that not happen.

ancestor: remove extra generator from lazyancestors.__iter__()

localrepo: fix a mixmatched arg name in createrepository() docstring Differential Revision: https://phab.mercurial-scm.org/D4548

error: ensure ProgrammingError message is always a str Since this error is internal-only and a runtime error, let's give it a treatment that makes it behave identically when repr()d on both Python 2 and Python 3. Differential Revision: https://phab.mercurial-scm.org/D4545

py3: whitelist a test caught by the ratchet Differential Revision: https://phab.mercurial-scm.org/D4547

tests: handle Python 3 not quoting non-empty-directory error I assume this happens on Windows too, so I did the same regex on both versions of the output. The whole message printed by these aborts comes from Python, so if we want to exert control over the quoting here it'll be a bit of a pain. Differential Revision: https://phab.mercurial-scm.org/D4546

context: don't count deleted files as candidates for path conflicts in IMM This patch makes sure we don't consider the deleted files in our IMM wctx as potential conflicts while calculating paths conflicts. This fixes the bug demonstrated in previous patch. Differential Revision: https://phab.mercurial-scm.org/D4543

rebase: add tests showing patch conflict detection needs to be smarter in IMM This patch adds test which shows that you can't rebase a cset which removes a dir and adds a file of the same as that of dir as there are False positives path conflicts reported. I fixed the case when there is a file and we adds a dir of same name while removing the file, but missed testing the current case. Next patch will fix this. Differential Revision: https://phab.mercurial-scm.org/D4544

zsh_completion: add new and remove deprecated flags Differential Revision: https://phab.mercurial-scm.org/D4519

zsh_completion: update various arguments, descriptions, metavariables Addition of "=" means the flag must have an argument after it. Differential Revision: https://phab.mercurial-scm.org/D4518

setup: don't support py 3.5.0, 3.5.1, 3.5.2 because of bug in codecs codecs.escape_encode() raises SystemError if an empty bytestring is passed. We do that at some places in our code and because of this bug, things break. Therefore we can't support the mentioned version. The bug was fixed in 3.5.3, 3.6.0 beta 2. We can't support 3.6.0 anyway because of bug in formatting bytestrings. Link to the python bug: https://bugs.python.org/issue25270 Differential Revision: https://phab.mercurial-scm.org/D4475

util: update lrucachedict order during get() get() should have the same semantics as __getitem__ for item retrieval. Differential Revision: https://phab.mercurial-scm.org/D4506

util: lower water mark when removing nodes after cost limit reached See the inline comment for the reasoning here. This is a pretty common strategy for garbage collectors, other cache-like primtives. The performance impact is substantial: $ hg perflrucachedict --size 4 --gets 1000000 --sets 1000000 --mixed 1000000 --costlimit 100 ! inserts w/ cost limit ! wall 1.659181 comb 1.650000 user 1.650000 sys 0.000000 (best of 7) ! wall 1.722122 comb 1.720000 user 1.720000 sys 0.000000 (best of 6) ! mixed w/ cost limit ! wall 1.139955 comb 1.140000 user 1.140000 sys 0.000000 (best of 9) ! wall 1.182513 comb 1.180000 user 1.180000 sys 0.000000 (best of 9) $ hg perflrucachedict --size 1000 --gets 1000000 --sets 1000000 --mixed 1000000 --costlimit 10000 ! inserts ! wall 0.679546 comb 0.680000 user 0.680000 sys 0.000000 (best of 15) ! sets ! wall 0.825147 comb 0.830000 user 0.830000 sys 0.000000 (best of 13) ! inserts w/ cost limit ! wall 25.105273 comb 25.080000 user 25.080000 sys 0.000000 (best of 3) ! wall 1.724397 comb 1.720000 user 1.720000 sys 0.000000 (best of 6) ! mixed ! wall 0.807096 comb 0.810000 user 0.810000 sys 0.000000 (best of 13) ! mixed w/ cost limit ! wall 12.104470 comb 12.070000 user 12.070000 sys 0.000000 (best of 3) ! wall 1.190563 comb 1.190000 user 1.190000 sys 0.000000 (best of 9) $ hg perflrucachedict --size 1000 --gets 1000000 --sets 1000000 --mixed 1000000 --costlimit 10000 --mixedgetfreq 90 ! inserts ! wall 0.711177 comb 0.710000 user 0.710000 sys 0.000000 (best of 14) ! sets ! wall 0.846992 comb 0.850000 user 0.850000 sys 0.000000 (best of 12) ! inserts w/ cost limit ! wall 25.963028 comb 25.960000 user 25.960000 sys 0.000000 (best of 3) ! wall 2.184311 comb 2.180000 user 2.180000 sys 0.000000 (best of 5) ! mixed ! wall 0.728256 comb 0.730000 user 0.730000 sys 0.000000 (best of 14) ! mixed w/ cost limit ! wall 3.174256 comb 3.170000 user 3.170000 sys 0.000000 (best of 4) ! wall 0.773186 comb 0.770000 user 0.770000 sys 0.000000 (best of 13) $ hg perflrucachedict --size 100000 --gets 1000000 --sets 1000000 --mixed 1000000 --mixedgetfreq 90 --costlimit 5000000 ! gets ! wall 1.191368 comb 1.190000 user 1.190000 sys 0.000000 (best of 9) ! wall 1.195304 comb 1.190000 user 1.190000 sys 0.000000 (best of 9) ! inserts ! wall 0.950995 comb 0.950000 user 0.950000 sys 0.000000 (best of 11) ! inserts w/ cost limit ! wall 1.589732 comb 1.590000 user 1.590000 sys 0.000000 (best of 7) ! sets ! wall 1.094941 comb 1.100000 user 1.090000 sys 0.010000 (best of 9) ! mixed ! wall 0.936420 comb 0.940000 user 0.930000 sys 0.010000 (best of 10) ! mixed w/ cost limit ! wall 0.882780 comb 0.870000 user 0.870000 sys 0.000000 (best of 11) This puts us ~2x slower than caches without cost accounting. And for read-heavy workloads (the prime use cases for caches), performance is nearly identical. In the worst case (pure write workloads with cost accounting enabled), we're looking at ~1.5us per insert on large caches. That seems "fast enough." Differential Revision: https://phab.mercurial-scm.org/D4505

util: optimize cost auditing on insert Calling popoldest() on insert with cost auditing enabled introduces significant overhead. The primary reason for this overhead is that popoldest() needs to walk the linked list to find the first non-empty node. When we call popoldest() within a loop, this can become quadratic. The performance impact is more pronounced on caches with large capacities. This commit effectively inlines the popoldest() call into _enforcecostlimit(). By doing so, we only do the backwards walk to find the first empty node once. However, we still may still perform this work on insert when the cache is near cost capacity. So this is only a partial performance win. $ hg perflrucachedict --size 4 --gets 1000000 --sets 1000000 --mixed 1000000 --costlimit 100 ! gets w/ cost limit ! wall 0.598737 comb 0.590000 user 0.590000 sys 0.000000 (best of 17) ! inserts w/ cost limit ! wall 1.694282 comb 1.700000 user 1.700000 sys 0.000000 (best of 6) ! wall 1.659181 comb 1.650000 user 1.650000 sys 0.000000 (best of 7) ! mixed w/ cost limit ! wall 1.157655 comb 1.150000 user 1.150000 sys 0.000000 (best of 9) ! wall 1.139955 comb 1.140000 user 1.140000 sys 0.000000 (best of 9) $ hg perflrucachedict --size 1000 --gets 1000000 --sets 1000000 --mixed 1000000 --costlimit 10000 ! gets w/ cost limit ! wall 0.598526 comb 0.600000 user 0.600000 sys 0.000000 (best of 17) ! wall 0.601993 comb 0.600000 user 0.600000 sys 0.000000 (best of 17) ! inserts w/ cost limit ! wall 37.838315 comb 37.840000 user 37.840000 sys 0.000000 (best of 3) ! wall 25.105273 comb 25.080000 user 25.080000 sys 0.000000 (best of 3) ! mixed w/ cost limit ! wall 18.060198 comb 18.060000 user 18.060000 sys 0.000000 (best of 3) ! wall 12.104470 comb 12.070000 user 12.070000 sys 0.000000 (best of 3) $ hg perflrucachedict --size 1000 --gets 1000000 --sets 1000000 --mixed 1000000 --costlimit 10000 --mixedgetfreq 90 ! gets w/ cost limit ! wall 0.600024 comb 0.600000 user 0.600000 sys 0.000000 (best of 17) ! wall 0.614439 comb 0.620000 user 0.620000 sys 0.000000 (best of 17) ! inserts w/ cost limit ! wall 37.154547 comb 37.120000 user 37.120000 sys 0.000000 (best of 3) ! wall 25.963028 comb 25.960000 user 25.960000 sys 0.000000 (best of 3) ! mixed w/ cost limit ! wall 4.381602 comb 4.380000 user 4.370000 sys 0.010000 (best of 3) ! wall 3.174256 comb 3.170000 user 3.170000 sys 0.000000 (best of 4) Differential Revision: https://phab.mercurial-scm.org/D4504

util: teach lrucachedict to enforce a max total cost Now that lrucachedict entries can have a numeric cost associated with them and we can easily pop the oldest item in the cache, it now becomes relatively trivial to implement support for enforcing a high water mark on the total cost of items in the cache. This commit teaches lrucachedict instances to have a max cost associated with them. When items are inserted, we pop old items until enough "cost" frees up to make room for the new item. This feature is close to zero cost when not used (modulo the insertion regressed introduced by the previous commit): $ ./hg perflrucachedict --size 4 --gets 1000000 --sets 1000000 --mixed 1000000 ! gets ! wall 0.607444 comb 0.610000 user 0.610000 sys 0.000000 (best of 17) ! wall 0.601653 comb 0.600000 user 0.600000 sys 0.000000 (best of 17) ! inserts ! wall 0.678261 comb 0.680000 user 0.680000 sys 0.000000 (best of 14) ! wall 0.685042 comb 0.680000 user 0.680000 sys 0.000000 (best of 15) ! sets ! wall 0.808770 comb 0.800000 user 0.800000 sys 0.000000 (best of 13) ! wall 0.834241 comb 0.830000 user 0.830000 sys 0.000000 (best of 12) ! mixed ! wall 0.782441 comb 0.780000 user 0.780000 sys 0.000000 (best of 13) ! wall 0.803804 comb 0.800000 user 0.800000 sys 0.000000 (best of 13) $ hg perflrucachedict --size 1000 --gets 1000000 --sets 1000000 --mixed 1000000 ! init ! wall 0.006952 comb 0.010000 user 0.010000 sys 0.000000 (best of 418) ! gets ! wall 0.613350 comb 0.610000 user 0.610000 sys 0.000000 (best of 17) ! wall 0.617415 comb 0.620000 user 0.620000 sys 0.000000 (best of 17) ! inserts ! wall 0.701270 comb 0.700000 user 0.700000 sys 0.000000 (best of 15) ! wall 0.700516 comb 0.700000 user 0.700000 sys 0.000000 (best of 15) ! sets ! wall 0.825720 comb 0.830000 user 0.830000 sys 0.000000 (best of 13) ! wall 0.837946 comb 0.840000 user 0.830000 sys 0.010000 (best of 12) ! mixed ! wall 0.821644 comb 0.820000 user 0.820000 sys 0.000000 (best of 13) ! wall 0.850559 comb 0.850000 user 0.850000 sys 0.000000 (best of 12) I reckon the slight slowdown on insert is due to added if checks. For caches with total cost limiting enabled: $ hg perflrucachedict --size 4 --gets 1000000 --sets 1000000 --mixed 1000000 --costlimit 100 ! gets w/ cost limit ! wall 0.598737 comb 0.590000 user 0.590000 sys 0.000000 (best of 17) ! inserts w/ cost limit ! wall 1.694282 comb 1.700000 user 1.700000 sys 0.000000 (best of 6) ! mixed w/ cost limit ! wall 1.157655 comb 1.150000 user 1.150000 sys 0.000000 (best of 9) $ hg perflrucachedict --size 1000 --gets 1000000 --sets 1000000 --mixed 1000000 --costlimit 10000 ! gets w/ cost limit ! wall 0.598526 comb 0.600000 user 0.600000 sys 0.000000 (best of 17) ! inserts w/ cost limit ! wall 37.838315 comb 37.840000 user 37.840000 sys 0.000000 (best of 3) ! mixed w/ cost limit ! wall 18.060198 comb 18.060000 user 18.060000 sys 0.000000 (best of 3) $ hg perflrucachedict --size 1000 --gets 1000000 --sets 1000000 --mixed 1000000 --costlimit 10000 --mixedgetfreq 90 ! gets w/ cost limit ! wall 0.600024 comb 0.600000 user 0.600000 sys 0.000000 (best of 17) ! inserts w/ cost limit ! wall 37.154547 comb 37.120000 user 37.120000 sys 0.000000 (best of 3) ! mixed w/ cost limit ! wall 4.381602 comb 4.380000 user 4.370000 sys 0.010000 (best of 3) The functions we're benchmarking are slightly different, which could move numbers by a few milliseconds. But the slowdown on insert is too great to be explained by that. The slowness is due to insert heavy operations needing to call popoldest() repeatedly when the cache is at capacity. The next commit will address this. Differential Revision: https://phab.mercurial-scm.org/D4503

util: allow lrucachedict to track cost of entries Currently, lrucachedict allows tracking of arbitrary items with the only limit being the total number of items in the cache. Caches can be a lot more useful when they are bound by the size of the items in them rather than the number of elements in the cache. In preparation for teaching lrucachedict to enforce a max size of cached items, we teach lrucachedict to optionally associate a numeric cost value with each node. We purposefully let the caller define their own cost for nodes. This does introduce some overhead. Most of it comes from __setitem__, since that function now calls into insert(), thus introducing Python function call overhead. $ hg perflrucachedict --size 4 --gets 1000000 --sets 1000000 --mixed 1000000 ! gets ! wall 0.599552 comb 0.600000 user 0.600000 sys 0.000000 (best of 17) ! wall 0.614643 comb 0.610000 user 0.610000 sys 0.000000 (best of 17) ! inserts ! <not available> ! wall 0.655817 comb 0.650000 user 0.650000 sys 0.000000 (best of 16) ! sets ! wall 0.540448 comb 0.540000 user 0.540000 sys 0.000000 (best of 18) ! wall 0.805644 comb 0.810000 user 0.810000 sys 0.000000 (best of 13) ! mixed ! wall 0.651556 comb 0.660000 user 0.660000 sys 0.000000 (best of 15) ! wall 0.781357 comb 0.780000 user 0.780000 sys 0.000000 (best of 13) $ hg perflrucachedict --size 1000 --gets 1000000 --sets 1000000 --mixed 1000000 ! gets ! wall 0.621014 comb 0.620000 user 0.620000 sys 0.000000 (best of 16) ! wall 0.615146 comb 0.620000 user 0.620000 sys 0.000000 (best of 17) ! inserts ! <not available> ! wall 0.698115 comb 0.700000 user 0.700000 sys 0.000000 (best of 15) ! sets ! wall 0.560247 comb 0.560000 user 0.560000 sys 0.000000 (best of 18) ! wall 0.832495 comb 0.830000 user 0.830000 sys 0.000000 (best of 12) ! mixed ! wall 0.686172 comb 0.680000 user 0.680000 sys 0.000000 (best of 15) ! wall 0.841359 comb 0.840000 user 0.840000 sys 0.000000 (best of 12) We're still under 1us per insert, which seems like reasonable performance for a cache. If we comment out updating of self.totalcost during insert(), performance of insert() is identical to __setitem__ before. However, I don't want to make total cost evaluation lazy because it has significant performance implications for when we need to evaluate the total cost at mutation time (it requires a cache traversal, which could be expensive for large caches). Differential Revision: https://phab.mercurial-scm.org/D4502

util: add a popoldest() method to lrucachedict This allows consumers to prune the oldest item from the cache. This could be useful for e.g. a consumer that wishes for the size of items tracked by the cache to remain under a high water mark. Differential Revision: https://phab.mercurial-scm.org/D4501

util: ability to change capacity when copying lrucachedict This will allow us to easily replace an lrucachedict with one with a higher or lower capacity as consumers deem necessary. IMO it is easier to just create a new cache instance than to muck with the capacity of an existing cache. Mutating an existing cache's capacity feels more prone to bugs. Differential Revision: https://phab.mercurial-scm.org/D4500

util: make capacity a public attribute on lrucachedict So others can query it. Useful for operations that may want to verify the cache has capacity for N items before it performs an operation that may cause cache eviction. Differential Revision: https://phab.mercurial-scm.org/D4499

util: properly copy lrucachedict instances Previously, copy() only worked if the cache was full. We teach copy() to only copy defined nodes. Differential Revision: https://phab.mercurial-scm.org/D4498

tests: rewrite test-lrucachedict.py to use unittest This makes the code so much easier to test and debug. Along the way, I discovered a bug in copy(), which I kind of added test coverage for. Differential Revision: https://phab.mercurial-scm.org/D4497

wireprotov2peer: stream decoded responses Previously, wire protocol version 2 would buffer all response data. Only once all data was received did we CBOR decode it and resolve the future associated with the command. This was obviously not desirable. In future commits that introduce large response payloads, this caused significant memory bloat and slowed down client operations due to waiting on the server. This commit refactors the response handling code so that response data can be streamed. Command response objects now contain a buffered CBOR decoder. As new data arrives, it is fed into the decoder. Decoded objects are made available to the generator as they are decoded. Because there is a separate thread processing incoming frames and feeding data into the response object, there is the potential for race conditions when mutating response objects. So a lock has been added to guard access to critical state variables. Because the generator emitting decoded objects needs to wait on those objects to become available, we've added an Event for the generator to wait on so it doesn't busy loop. This does mean there is the potential for deadlocks. And I'm pretty sure they can occur in some scenarios. We already have a handful of TODOs around this. But I've added some more. Fixing this will likely require moving the background thread receiving frames into clienthandler. We likely would have done this anyway when implementing the client bits for the SSH transport. Test output changes because the initial CBOR map holding the overall response state is now always handled internally by the response object. Differential Revision: https://phab.mercurial-scm.org/D4474

wireprotoframing: buffer emitted data to reduce frame count An upcoming commit introduces a wire protocol command that can emit hundreds of thousands of small objects. Without a buffering layer, we would emit a single, small frame for every object. Performance profiling revealed this to be a source of significant overhead for both client and server. This commit introduces a very crude buffering layer so that we emit fewer, bigger frames in such a scenario. This code will likely get rewritten in the future to be part of the streams API, as we'll need a similar strategy for compressing data. I don't want to think about it too much at the moment though. server before: user 32.500+0.000 sys 1.160+0.000 after: user 20.230+0.010 sys 0.180+0.000 client before: user 133.400+0.000 sys 93.120+0.000 after: user 68.370+0.000 sys 32.950+0.000 This appears to indicate we have significant overhead in the frame processing code on both client and server. It might be worth profiling that at some point... Differential Revision: https://phab.mercurial-scm.org/D4473

wireprotov2: implement commands as a generator of objects Previously, wire protocol version 2 inherited version 1's model of having separate types to represent the results of different wire protocol commands. As I implemented more powerful commands in future commits, I found I was using a common pattern of returning a special type to hold a generator. This meant the command function required a closure to do most of the work. That made logic flow more difficult to follow. I also noticed that many commands were effectively a sequence of objects to be CBOR encoded. I think it makes sense to define version 2 commands as generators. This way, commands can simply emit the data structures they wish to send to the client. This eliminates the need for a closure in command functions and removes encoding from the bodies of commands. As part of this commit, the handling of response objects has been moved into the serverreactor class. This puts the reactor in the driver's seat with regards to CBOR encoding and error handling. Having error handling in the function that emits frames is particularly important because exceptions in that function can lead to things getting in a bad state: I'm fairly certain that uncaught exceptions in the frame generator were causing deadlocks. I also introduced a dedicated error type for explicit error reporting in command handlers. This will be used in subsequent commits. There's still a bit of work to be done here, especially around formalizing the error handling "protocol." I've added yet another TODO to track this so we don't forget. Test output changed because we're using generators and no longer know we are at the end of the data until we hit the end of the generator. This means we can't emit the end-of-stream flag until we've exhausted the generator. Hence the introduction of 0-sized end-of-stream frames. Differential Revision: https://phab.mercurial-scm.org/D4472

internals: extract frame-based protocol docs to own document wireprotocol.txt is quite long and difficult to digest. The frame-based protocol is effectively a standalone concept (and could even be used outside of Mercurial). So this commit extracts its docs to a standalone file. The first few paragraphs were rewritten as part of the extraction. Sections headers were adjusted accordingly. Existing referalls in wireprotocol.txt were updated to refer to the new doc / concept, which I've started referring to as `hgrpc`. I'm on the fence as to whether to move the HTTP and SSH transport details to the new doc as well. For now, I'm leaving them in wireprotocol.txt. Differential Revision: https://phab.mercurial-scm.org/D4443

narrow: remove hack to write narrowspec to shared .hg directory AFAIK, we no longer need it since the narrowspec file was move to the store directory in 576eef1ab43d, "narrow: move .hg/narrowspec to .hg/store/narrowspec."

narrowspec: remove parseserverpatterns() which isn't used anymore Follows up 10a8472f6662, "narrow: drop support for remote expansion."

hg: write narrow patterns after repo creation Now that hg.clone() knows when a narrow clone is requested, it makes sense to have it update the narrow patterns for the repo soon after the repo is created, before any exchange occurs. Previously, the narrow extension was monkeypatching an exchange function to do this. The old code is redundant and has been removed. Differential Revision: https://phab.mercurial-scm.org/D4541

narrow: don't wrap exchange.pull() during clone The wrapped version was setting up the narrow repo requirement when a narrow clone was requested. Previous commits taught hg.clone() and repo creation to add the narrow requirement when a narrow clone was requested. So this requirement should already be set up for us and this code is no longer necessary. Differential Revision: https://phab.mercurial-scm.org/D4540

exchange: support defining narrow file patterns for pull This commit teaches exchange.pull() about the desire to perform a narrow file pull. We simply pass include and exclude patterns to the function. The values are validated and stored on the pulloperation instance. hg.clone() has been taught to pass these arguments to exchange.pull(). If the arguments are not passed to exchange.pull(), the active narrow patterns from the repository will automatically be used. We /could/ always use the narrow patterns from the repo. However, allowing explicit values to be passed in allows us to perform data fetching that doesn't necessarily align with the repo configuration. This provides more flexibility. Differential Revision: https://phab.mercurial-scm.org/D4539

commands: pass include and exclude options to hg.clone() These arguments are defined by the narrow extension. Let's teach core to recognize them so we can delete some code from the narrow extension and start to exercise the in-core code for performing a narrow clone. We have no way of easily testing it, but this change should result in .hg/requires having the narrow requirement from the time the file is written rather than added as part of pull. We'll confirm this when we delete some monkeypatched functions from the narrow extension in later commits. Test output changed because hg.clone() is now receiving patterns and validation of those values is occurring sooner, before the exchange code runs and prints the message that was deleted. Differential Revision: https://phab.mercurial-scm.org/D4538

localrepo: add requirement when narrow files creation option present The previous commit taught hg.clone() to define a creation option when file include or exclude patterns are passed. This commit teaches the new repo creation code to convert that creation option into a repository requirement. While not yet used by the narrow extension, the eventual side-effect of this change is that newly-created repositories will have the narrow requirement from their creation onset. Currently, the requirement is added to the repo at exchange.pull() time via a wrapped function in the narrow extension. Differential Revision: https://phab.mercurial-scm.org/D4537

hg: recognize include and exclude patterns when cloning This commit teaches clone() to accept arguments defining file patterns to clone. This is the first step in teaching core code about the existence of a narrow clone. Right now, we only perform validation of the arguments and pass additional options into createopts to influence repository creation. Nothing of consequence happens with that creation option yet, however. For now, arbitrary restrictions exist, such as not allowing patterns for shared repos and disabling local copies when patterns are defined. We can potentially lift these restrictions in the future once partial clone/storage support is more flushed out. I figure it is best to reduce the surface area for bugs for the time being. It may seem weird to prefix these arguments with "store." However, clone is effectively pull + update and file patterns could apply to both the store and the working directory. The prefix is there to disambiguate in the future when this function may want to use different sets of patterns for the store and working directory. Differential Revision: https://phab.mercurial-scm.org/D4536

hg: allow extra arguments to be passed to repo creation (API) Currently, repository creation is influenced by consulting the ui instance and turning config options into requirements. This means that in order to influence repository creation, you need to define and set a config option and that the option must translate to a requirement stored in the .hg/requires file. This commit introduces a new mechanism to influence repository creation. hg.repository() and hg.peer() have been taught to receive a new optional argument defining extra options to apply to repository creation. This value is passed along to the various instance() functions and can be used to influence repository creation. This will allow us to pass rich data directly to repository creation without having to go through the config layer. It also allows us to be more explicit about the features requested during repository creation and provides a natural point to detect unhandled options influencing repository creation. The new code detects when unknown creation options are present and aborts in that case. .. api:: options can now be passed to influence repository creation The various instance() functions to spawn new peers or repository instances now receive a ``createopts`` argument that can be a dict defining additional options to influence repository creation. localrepo.newreporequirements() also receives this argument. Differential Revision: https://phab.mercurial-scm.org/D4535

localrepo: move repo creation logic out of localrepository.__init__ (API) It has long bothered me that local repository creation is handled as part of localrepository.__init__. Upcoming changes I want to make around how repositories are initialized and instantiated will make the continued existence of repository creation code in localrepository.__init__ even more awkward. localrepository instances are almost never constructed directly: instead, callers are supposed to go through hg.repository() to obtain a handle on a repository. And hg.repository() calls localrepo.instance() to return a new repo instance. This commit teaches localrepo.instance() to handle the create=True logic. Most of the code for repo construction has been moved to a standalone function. This allows extensions to monkeypatch the function to further customize freshly-created repositories. A few calls to localrepo.localrepository.__init__ that were passing create=True were converted to call localrepo.instance(). .. api:: local repo creation moved out of constructor ``localrepo.localrepository.__init__`` no longer accepts a ``create`` argument to create a new repository. New repository creation is now performed as part of ``localrepo.instance()`` and the bulk of the work is performed by ``localrepo.createrepository()``. Differential Revision: https://phab.mercurial-scm.org/D4534

localrepo: pass ui to newreporequirements() (API) newreporequirements() is called as part of creating a new repository. It doesn't make much sense for it to receive a repo instance as part of determining what requirements for new repos should be. .. api:: localrepo.newreporequirements() receives a ui instead of a repo Differential Revision: https://phab.mercurial-scm.org/D4533

narrow: set opts['narrow'] instead of local variable This will allow the command function in core to infer the presence of the option without duplicating logic. Differential Revision: https://phab.mercurial-scm.org/D4532

narrow: drop support for remote expansion (BC) Previous patches to validate narrow patterns accidentically dropped support for the include: syntax that allows patterns to be expanded from a remote. This feature was never implemented in core and is only implemented on Google's custom server. Per @martinvonz's review comment in D4522, it is OK to drop this feature since it isn't used. The concept of this feature does seem useful. I anticipate it making a comeback some day in some shape or form. But for now, let's jettison the dead code. Differential Revision: https://phab.mercurial-scm.org/D4530

fastannotate: use repo.local() This is the proper way to check whether we're dealing with a local repository, since extensions should be coding to an interface and not testing for exact types. Differential Revision: https://phab.mercurial-scm.org/D4542

tests: drop extra "file:" prefix from paths in narrow test It looks like these were added by mistake in f4d4bd8c8911 (narrow: add a --narrowspec flag to clone command, 2018-08-08). Differential Revision: https://phab.mercurial-scm.org/D4531

narrow: validate spec files are well-formed during clone (BC) Previously, specfiles would get read then normalized. We want specfiles to be normalized on read so there is no confusion about what the format of specfiles should be. This commit validates the parsed result of --specfile. If entries aren't prefixed, an error is raised. Previously, validation would occur at exchange time, hence why we dropped a line of test output related to server iteraction. Differential Revision: https://phab.mercurial-scm.org/D4526

narrow: validate patterns on incoming bundle2 part The remote data is untrusted and needs to be validated for pattern conformance. Differential Revision: https://phab.mercurial-scm.org/D4525

narrowspec: validate patterns when loading and saving spec file Patterns should be normalized and validated before being passed into narrowspec.save(). Let's assert that by checking immediately before writing the narrow spec file. And let's assert that patterns loaded from the spec file also conform. Differential Revision: https://phab.mercurial-scm.org/D4524

ancestor: use heapreplace() in place of heappop/heappush() This should be slightly faster. Overall perfancestors result:: cpython nginx mercurial ------------- ---------------- ---------------- ---------------- b6db2e80a9ce^ 0.103461 0.006303 0.035716 8eb2145ff0fb 0.192307 (x1.86) 0.012115 (x1.92) 0.052135 (x1.46) this patch 0.139986 (x1.35) 0.006389 (x1.01) 0.037176 (x1.04)

ancestor: rename local aliases of heapq functions in _lazyancestorsiter() The original names no longer look pretty. Just call them as heap*() instead.

ancestor: optimize _lazyancestorsiter() for contiguous chains If there's no revision between p1 and current, p1 must be the next revision to visit. In this case, we can get around the overhead of heappop/push operations. Note that this is faster than using heapreplace(). 'current - p1 == 1' could be generalized as 'all(r not in seen for r in xrange(p1, current)', but Python is too slow to do such thing.

ancestor: unroll loop of parents in _lazyancestorsiter() This change itself isn't major performance win, but it helps optimizing the visit loop for contiguous chains. See the next patch.

ancestor: return early from _lazyancestorsiter() when reached to stoprev There's no need to empty the heap.

ancestor: remove alias of initrevs from _lazyancestorsiter() It's just redundant and less comprehensible.

narrow: validate patterns returned by expandnarrow Remotes could supply malicious or invalid patterns. We should validate them as soon as possible. Differential Revision: https://phab.mercurial-scm.org/D4523

narrowspec: limit patterns to path: and rootfilesin: (BC) Some matcher patterns are computationally expensive and may even have security issues (e.g. evaluating some file sets). For these reasons, we want to limit the types of matcher patterns that can be used in narrow specs and by command line arguments used for defining narrow specs. This commit teaches ``narrowspec.parsepatterns()`` to validate the pattern types against "safe" patterns. Surprisingly, no existing tests broke. So tests for the feature have been added. We also added a function to validate a patterns data structure. This will be used in future commits. Differential Revision: https://phab.mercurial-scm.org/D4522

narrow: mark wire proto capability names experimental and versioned We already plan to add a "widen" wire protocol command to the "narrow" capability, so let's version the capabilities as "exp-narrow-1" and "exp-ellipses-1". When we add the "widen" command, we will then add a "exp-narrow-2" capability to indicate support for that command. Differential Revision: https://phab.mercurial-scm.org/D4529

narrow: move wire proto capabilities to narrowwirepeer These are not bundle2 capabilities (they just happened to share the name "narrow"), so they seem to belong with the wirepeer overrides. Differential Revision: https://phab.mercurial-scm.org/D4528