match: add missing "return set()", add FIXME to test to doc a bug These were both brought up during the codereview of D4130. Differential Revision: https://phab.mercurial-scm.org/D4160

match: correct doc for _rootsdirsandparents after 5a7df82de142 Differential Revision: https://phab.mercurial-scm.org/D4159

dirstate: use visitchildrenset in traverse This speeds up `hg status` a fair amount when there is a very large directory and narrow is in use. Timing numbers according to command: hyperfine --warmup 1 'hg status' HGRCPATH points to a file with the following contents: [extensions] narrow = mozilla-unified (called m-u below) was at revision #468856. regular hash: eb39298e432d treemanifests hash: 0553b7f29eaf large-dir-repo (called l-d-r below) was generated with the following script: #!/bin/bash hg init large-dir-repo mkdir -p large-dir-repo/third_party/rust/log touch large-dir-repo/third_party/rust/log/foo.txt for i in $(seq 1 30000); do d=$(mktemp -d large-dir-repo/third_party/XXXXXXXXX) touch $d/file.txt done hg -R large-dir-repo ci -Am 'rev0' --user test --date '0 0' for repos that use narrow, the narrowspec was this: [includes] rootfilesin:third_party/rust/log [excludes] This narrowspec was chosen due to the size of the third_party/rust directory; this directory was *not* modified in revision #468856 in mozilla-unified. Importantly, when using narrow, these repos had everything checked out (in the case of large-dir-repo, that means all 30,001 directories), *before* adding the narrowspec. This is to simulate the behavior when using a virtual filesystem that shows everything for the user even if they haven't added it to the narrowspec yet. This is not a supported configuration, and `hg update` will not really do the "correct" thing, but non-mutating commands should behave correctly. There are two repos below that do not follow the setup above, 'citc1' and 'citc2', which are using a virtual filesystem and can not be reproduced upstream; these numbers are here mostly to indicate that these performance improvements are not hypothetical, and show the benefits we're hoping to achieve on our real workloads. 'citc1' is closest to large-dir-repo with one of our pathological cases, 'citc2' is an arbitrary repo and closer to "average". I'm not claiming anything less than a 5% speed win as improvements due to this change; these are probably eiter measurement artifacts or constant time improvements. The numbers that aren't changing are shown primarily to prove that this doesn't make anything worse in any case I plan on testing during this series. 'before' is hg from commit c83ad576. 'N' indicates narrow in use, 'T' indicates treemanifest in use. hg status: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 2.284 s +- 0.022 s | 2.274 s +- 0.021 s | 99.6% m-u | | x | 2.289 s +- 0.008 s | 2.284 s +- 0.028 s | 99.8% m-u | x | | 430.8 ms +- 3.1 ms | 424.5 ms +- 3.2 ms | 98.5% m-u | x | x | 429.8 ms +- 2.5 ms | 425.8 ms +- 3.7 ms | 99.1% l-d-r | | | 681.3 ms +- 5.5 ms | 689.6 ms +- 8.0 ms | 101.2% l-d-r | | x | 666.8 ms +- 21.8 ms | 672.5 ms +- 14.9 ms | 100.9% l-d-r | x | | 282.6 ms +- 1.8 ms | 203.0 ms +- 1.2 ms | 71.8% <-- l-d-r | x | x | 275.2 ms +- 3.9 ms | 199.3 ms +- 3.5 ms | 72.4% <-- citc1 | x | x | 1.023 s +- 0.011 s | 398.6 ms +- 9.2 ms | 39.0% <-- citc2 | x | x | 297.9 ms +- 4.4 ms | 289.6 ms +- 4.2 ms | 97.2% hg status --change .: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 478.2 ms +- 2.0 ms | 476.9 ms +- 3.7 ms | 99.7% m-u | | x | 169.5 ms +- 2.7 ms | 169.5 ms +- 2.5 ms | 100.0% m-u | x | | 477.0 ms +- 2.4 ms | 476.1 ms +- 1.4 ms | 99.8% m-u | x | x | 124.7 ms +- 1.9 ms | 124.2 ms +- 3.3 ms | 99.6% l-d-r | | | 97.4 ms +- 1.2 ms | 96.5 ms +- 1.2 ms | 99.1% l-d-r | | x | 4.778 s +- 0.018 s | 4.774 s +- 0.011 s | 99.9% l-d-r | x | | 99.9 ms +- 1.1 ms | 98.8 ms +- 1.3 ms | 98.9% l-d-r | x | x | 848.7 ms +- 7.1 ms | 849.4 ms +- 6.5 ms | 100.1% citc1 | x | x | 4.250 s +- 0.051 s | 4.283 s +- 0.042 s | 100.8% citc2 | x | x | 341.5 ms +- 4.7 ms | 341.5 ms +- 4.1 ms | 100.0% hg update $rev^; hg update $rev: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 4.357 s +- 0.032 s | 4.312 s +- 0.093 s | 99.0% m-u | | x | 3.599 s +- 0.061 s | 3.592 s +- 0.071 s | 99.8% m-u | x | | 1.815 s +- 0.012 s | 1.816 s +- 0.013 s | 100.1% m-u | x | x | 1.110 s +- 0.009 s | 1.106 s +- 0.005 s | 99.6% l-d-r | | | 527.1 ms +- 7.8 ms | 523.3 ms +- 6.5 ms | 99.3% l-d-r | | x | 8.835 s +- 0.067 s | 8.825 s +- 0.064 s | 99.9% l-d-r | x | | 313.0 ms +- 2.2 ms | 312.1 ms +- 1.2 ms | 99.7% l-d-r | x | x | 1.780 s +- 0.011 s | 1.799 s +- 0.013 s | 101.1% citc1 | x | x | 6.825 s +- 0.262 s | 6.707 s +- 0.353 s | 98.3% citc2 | x | x | 776.4 ms +- 4.5 ms | 781.3 ms +- 6.3 ms | 100.6% hg diff: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+------------------------+-----------------------+------------ m-u | | | 1.519 s +- 0.015 s | 1.525 s +- 0.017 s | 100.4% m-u | | x | 1.512 s +- 0.010 s | 1.517 s +- 0.027 s | 100.3% m-u | x | | 420.0 ms +- 3.2 ms | 417.1 ms +- 1.9 ms | 99.3% m-u | x | x | 415.0 ms +- 3.8 ms | 415.7 ms +- 2.7 ms | 100.2% l-d-r | | | 220.8 ms +- 4.0 ms | 220.8 ms +- 3.7 ms | 100.0% l-d-r | | x | 216.6 ms +- 7.5 ms | 211.4 ms +- 2.1 ms | 97.6% l-d-r | x | | 111.9 ms +- 1.8 ms | 112.0 ms +- 1.5 ms | 100.1% l-d-r | x | x | 111.4 ms +- 1.4 ms | 110.2 ms +- 1.0 ms | 98.9% citc1 | x | x | 268.7 ms +- 2.3 ms | 269.6 ms +- 2.8 ms | 100.3% citc2 | x | x | 273.5 ms +- 5.5 ms | 273.9 ms +- 3.7 ms | 100.1% hg diff -c .: repo | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before ------+---+---+--------------------------+-----------------------+---------- m-u | | | 497.1 ms +- 1.4 ms | 500.1 ms +- 2.4 ms | 100.6% m-u | | x | 195.3 ms +- 13.2 ms | 191.6 ms +- 3.0 ms | 98.1% m-u | x | | 476.8 ms +- 1.9 ms | 476.7 ms +- 2.3 ms | 100.0% m-u | x | x | 122.8 ms +- 2.1 ms | 122.9 ms +- 2.0 ms | 100.1% l-d-r | | | 99.3 ms +- 2.3 ms | 98.8 ms +- 1.7 ms | 99.5% l-d-r | | x | 4.875 s +- 0.041 s | 4.847 s +- 0.038 s | 99.4% l-d-r | x | | 98.5 ms +- 1.2 ms | 98.9 ms +- 1.3 ms | 100.4% l-d-r | x | x | 864.6 ms +- 7.4 ms | 855.4 ms +- 6.6 ms | 98.9% citc1 | x | x | 4.505 s +- 0.060 s | 4.466 s +- 0.036 s | 99.1% citc2 | x | x | 368.0 ms +- 4.0 ms | 365.5 ms +- 6.3 ms | 99.3% Differential Revision: https://phab.mercurial-scm.org/D4131

match: add visitchildrenset complement to visitdir `visitdir(d)` lets a caller query whether the directory is part of the matcher. It can receive a response of 'all' (yes, and all children, you can stop calling visitdir now), False (no, and no children either), or True (yes, either something in this directory or a child is part of the matcher). `visitchildrenset(d)` augments that by instead of returning True, it returns a list of items to actually investigate. With this, code can be modified from: for f in self.all_items: if match.visitdir(self.dir + '/' + f): <do stuff> to be: for f in self.all_items.intersect(match.visitchildrenset(self.dir)): <do stuff> use of this function can provide significant performance improvements, especially when using narrow (so that the matcher is much smaller than the stuff we see on disk) and/or treemanifests (so that we can avoid loading manifests for trees that aren't part of the matcher). Differential Revision: https://phab.mercurial-scm.org/D4130

includematcher: separate "parents" from "dirs" A future patch will make use of this separation so that we can make more intelligent decisions about what to investigate/load when the matcher is in use. Currently, even with this patch, we typically use the 'visitdir' call to identify if we can skip some directory, something along the lines of: for f in all_items: if match.visitdir(f): <do stuff> This can be slower than we'd like if there are a lot of items; it requires N calls to match.visitdir in the best case. Commonly, especially with 'narrow', we have a situation where we do some work for the directory, possibly just loading it from disk (when using treemanifests) and then check if we should be interacting with it at all, which can be a huge slowdown in some pathological cases. Differential Revision: https://phab.mercurial-scm.org/D4129

match: add tests for visitdir functionality There are a few cases that we could have done better with some additional logic; I tried to annotate these when I noticed them, but may have missed some. The tests are not exhaustive; there are certainly some patterns that I didn't test well, and many that I didn't test at all. The primary motivation was to get coverage on visitdir so that I can cover identical cases in a similar method I'm working on, to make sure that this new method behaves the same (or better). Differential Revision: https://phab.mercurial-scm.org/D4128

mergetool: warn if ui.merge points to nonexistent tool This adds a warning when ui.merge is configured but points to an executable that doesn't exist. It gets printed once per fail, but that seems to be how our other warnings about merge tools are reported. Differential Revision: https://phab.mercurial-scm.org/D3975

tests: demonstrate that no requested merge tool is ignored if missing If you explicitly configure a merge tool, it seems wrong that we don't even warn if we can't find it. This patch adds a test case that demonstrates that. Differential Revision: https://phab.mercurial-scm.org/D3974

fix: correctly set wdirwritten given that the dict item is deleted Differential Revision: https://phab.mercurial-scm.org/D4146

fix: pull out flag definitions to make them re-usable from extensions This makes it cleaner to implement fix-related commands with additional functionality while sharing some flags with the core implementation. Differential Revision: https://phab.mercurial-scm.org/D4145

templatekw: copy {author} to {user} and document {author} as an alias In other places including "log -Tjson" and revset, "user" is the canonical name. Let's standardize it. This is a part of the name unification of the Generic Templating Plan. https://www.mercurial-scm.org/wiki/GenericTemplatingPlan#Dictionary

templates: rename "user" to "luser" defined in default map file (API) "user" will be shadowed by the {user} keyword to be added by the next patch. I think the naming of template fields is a sort of an internal API, so this patch is flagged as an API change. .. api:: Rewrite ``{user}`` to ``{luser}`` in log templates inherited from map-cmdline.default.

grep: add MULTIREV support to --allfiles flag This patch facilitates passing multiple revisions with all-files flag. It's assumed that if you are passing multiple revisions to --allfiles, you want hits from all of them. Differential Revision: https://phab.mercurial-scm.org/D3976

phabricator: convert description into local The description from conduit is a unicode. Differential Revision: https://phab.mercurial-scm.org/D3980

index: move index_clearcaches() further down I want to add a call from it to a new function (nt_dealloc) that will be inserted below its current position. Differential Revision: https://phab.mercurial-scm.org/D4117

index: move all "nt_*" functions to one place Differential Revision: https://phab.mercurial-scm.org/D4116

index: rename "nt_*(indexObject *self,...)" functions to "index_*" These functions do something with the nodetree, but they're less generic and won't make sense as methods on the nodetree when it becomes a Python type. Differential Revision: https://phab.mercurial-scm.org/D4115

index: split up nt_init() in two I'd like to make nt_init() take a pointer to a nodetree to initialize, but it currently also allocates the nodetree. This patch prepares for that change by making nt_init() be about initializing an existing node tree and by creating a new index_init_nt() that creates the nodetree. Differential Revision: https://phab.mercurial-scm.org/D4114

index: make most "nt_*" functions take a nodetree Now that the nodetree has a pointer to the index, we can pass the nodtree instead of the index. There are few "nt_*" functions left after this. I'll deal with them soon. Differential Revision: https://phab.mercurial-scm.org/D4113

index: add pointer from nodetree back to index This is always a cycle right now, but it will not be for the nodetree instances I'm planning to add later (see earlier patch). Differential Revision: https://phab.mercurial-scm.org/D4112

index: remove side-effect from failed nt_new() As pointed out by Yuya in the review of D4108, if realloc() fails, we would end up with an invalid nodetree instance (with nt->nodes set to NULL), which means that if it was later accessed again it would likely segfault. It's probably unlikely that much else happens in the process if it ran out memory, but we should of course do our best to handle it. This patch makes it so we don't update the nodetree in this case. Differential Revision: https://phab.mercurial-scm.org/D4154

index: remove side-effect from failed nt_init() As pointed out by Yuya in the review of D4108, if we run into the "overflow in nt_init" case (which I think normally happens only in repos with at least 2^26=64Mi revisions), we would leave the node tree half-initialized. Differential Revision: https://phab.mercurial-scm.org/D4153

index: use PyMem_Free() to free nodeetree instance As Yuya pointed out in the review of D4108, PyMem_Malloc() and PyMem_Free() should be paired. IIUC, PyMem_Malloc() may use a different allocator than malloc(), so using free() with a pointer from PyMem_Malloc() may be very wrong. Differential Revision: https://phab.mercurial-scm.org/D4152

linelog: fix infinite loop vulnerability Checking `len(lines)` is not a great way of detecting infinite loops, as demonstrated in the added test. Therefore check instruction count instead. The original C implementation does not have this problem. There are a few other places where the C implementation enforces more strictly, like `a1 <= a2`, `b1 <= b2`, `rev > 0`. But they are optional. Test Plan: Add a test. The old code forces the test to time out. Differential Revision: https://phab.mercurial-scm.org/D4151

tests: fix bytes/str issues in run-tests.py caught by python3 Differential Revision: https://phab.mercurial-scm.org/D4143

changegroup: assign to proper attribute 0548f696795b accidentally assigned to self.clrevtolocalrev instead of self._clrevtolocalrev. Surprisingly, no tests failed as a result of this mistake. Curious. Differential Revision: https://phab.mercurial-scm.org/D4144

absorb: remove sf alias for command I'm not even sure what it is supposed to stand for. Differential Revision: https://phab.mercurial-scm.org/D4126

hgweb: catch ParseError that's raised by revset.match() Some queries, like the demonstrated "first(::)", fail earlier than we call mfunc(), and that results in a "500 Internal Server Error". To prevent it, revset.match() also needs to be in a try-except block.

templatekw: deprecate old-style template keyword function (API) .. api:: `f(**kwargs)` style template keyword function is deprecated. Switch to new `(context, mapping)` API by declaring resource requirements. The new-style API will be the default in Mercurial 4.9. See registrar.templatekeyword for details.

hgweb: mark all lambda template keywords as new-style function This is just a temporary workaround, and will be removed in Mercurial 4.9.

hgweb: use registrar to add "motd" template keyword This prepares for deprecation of old-style keyword functions.

fileset: load core predicates directly to symbols dict We no longer have any side effect in loadpredicate().

fileset: turn on listclean conditionally This is just a micro optimization.

changegroup: always use the treemanifest-enabled version of _packmanifests() It works for flat manifests too. We just cannot use cg1 or cg2 if we have subdirectory manifests. Differential Revision: https://phab.mercurial-scm.org/D4124

linelog: add replacelines_vec for fastannotate # no-check-commit because we're conforming to an existing interface Differential Revision: https://phab.mercurial-scm.org/D3993

absorb: drop wrapper around the amend command We can reinstate this later if we want. Differential Revision: https://phab.mercurial-scm.org/D3992

absorb: note some TODOs from the code review Differential Revision: https://phab.mercurial-scm.org/D4047

absorb: use ui.debug() instead of open-coding it Differential Revision: https://phab.mercurial-scm.org/D4046

absorb: use pycompat to get xrange Differential Revision: https://phab.mercurial-scm.org/D4045

absorb: use set literal to avoid intermediate list Differential Revision: https://phab.mercurial-scm.org/D4044

absorb: avoid mutable default arg Differential Revision: https://phab.mercurial-scm.org/D4043

absorb: import extension from Facebook's hg-experimental absorb is a wicked-fast command to use blame information to automatically amend edits to the correct draft revision. Originally written by Jun Wu, this import is hgext3rd/absorb/__init__.py with: * the `testedwith` value changed * the linelog import updated * some missing configitems registered * some imports reordered per check-code.py * some missing __future__ imports added per check-code.py Differential Revision: https://phab.mercurial-scm.org/D3991

resolve: organize 'if confirm' conditionals Differential Revision: https://phab.mercurial-scm.org/D4123

index: pass only nodetree to nt_new() The function now only depends on the nodetree, not the index. Differential Revision: https://phab.mercurial-scm.org/D4111

index: drop now-redundant "nt" prefix of fields in nodetree struct Differential Revision: https://phab.mercurial-scm.org/D4110

index: move more fields onto nodetree type The fields moves are the ones that are not related to how the nodetree is used in the index and that will make sense for the new nodetree instance for a subset of the index that I'll add later. Differential Revision: https://phab.mercurial-scm.org/D4109

index: extract a type for the nodetree This is a first step towards exposing the nodetree as a Python type. Differential Revision: https://phab.mercurial-scm.org/D4108

index: make "nt_*" functions work on an initialized nodetree I want to be able to reuse these functions with another nodetree instance later (for disambiguating node prefix within a revset). That other nodetree instance won't want to be fully populated from the index, so this commit moves that part to the callers. Differential Revision: https://phab.mercurial-scm.org/D4107

changegroup: inline _packellipsischangegroup It now does nothing special. The logic is simple enough to inline in the 2 callers in narrow that need it. The changegroup generation APIs could probably be a bit simpler. But that's for another time. Differential Revision: https://phab.mercurial-scm.org/D4092

changegroup: move fullnodes into cgpacker And with this change, the narrow packer no longer defines any addition attributes on packer instances! Differential Revision: https://phab.mercurial-scm.org/D4091

changegroup: specify ellipses mode explicitly Currently, code throughout changegroup relies on the presence of self._full_nodes to enable ellipses mode. This is a very tenuous check. And the check may be wrong once we move _full_nodes into cgpacker. Let's capture the enabling of ellipses mode explicitly as a constructor argument and as an instance variable. We could probably derive ellipses mode by presence of other variables. But for now, this explicit approach seems simplest since it is most compatible with existing code. Differential Revision: https://phab.mercurial-scm.org/D4090

changegroup: pass ellipsis roots into cgpacker constructor And rename the internal variable to conform with naming conventions. Differential Revision: https://phab.mercurial-scm.org/D4089

changegroup: move revision maps to cgpacker And remove the underscores so the variables conform to our naming convention. The logic in _close() should be the only thing warranting scrutiny during review. Differential Revision: https://phab.mercurial-scm.org/D4088

changegroup: move changelogdone into cgpacker Looking at what it is used for, it feels like there is a better way to implement all this. So recording a TODO to track that. Differential Revision: https://phab.mercurial-scm.org/D4087

changegroup: declare shallow flag in constructor Thus begins the process of better formalizing ellipses and shallow changegroup generation mode so it is tracked by cgpacker at construction time instead of bolted on after the fact by a wrapper function. Differential Revision: https://phab.mercurial-scm.org/D4086

changegroup: make some packer attributes private These methods and attributes are low level and should not be called or outside outside of instances. Indicate as such through naming. Differential Revision: https://phab.mercurial-scm.org/D4085

changegroup: rename cg1packer to cgpacker There is now only a single class. We don't need to encode the version in its name since the version is a lie. Differential Revision: https://phab.mercurial-scm.org/D4084

changegroup: control delta parent behavior via constructor The last remaining override on cg2packer related to parent delta computation. We pass a parameter to the constructor to control whether to delta against the previous revision and we inline all parent delta logic into a single function. With this change, cg2packer is empty, so it has been deleted. Differential Revision: https://phab.mercurial-scm.org/D4083

changegroup: control reordering via constructor argument cg2packer.__init__ exists just to override self._reorder. Let's parameterize this behavior via an argument to the parent's __init__. The logic for self._reorder is kinda wonky. None is used as a special value and the value should be None in some situations. It is probably worth rewriting this logic to make behavior more explicit. This will likely happen as part of future work to control the delta generation process that I have planned. Differential Revision: https://phab.mercurial-scm.org/D4082

changegroup: consolidate tree manifests sending into cg1packer Previously, we overrode a method to control how manifests were serialized. This method was redefined on cg3packer to send tree manifests. This commit moves the tree manifests sending variation to cg1packer and teaches the cgpacker constructor to control which version to use. After these changes, cg3packer was empty. So it has been removed. Differential Revision: https://phab.mercurial-scm.org/D4081

changegroup: pass end of manifests marker into constructor cg3 inserts a custom marker in the stream once all manifests have been transferred. This is currently abstracted out by overriding a method. Let's pass the end of manifests marker in as an argument to avoid the extra method. Differential Revision: https://phab.mercurial-scm.org/D4080

changegroup: pass function to build delta header into constructor Previously, the delta header struct format was defined on each class and each class had a separate function for building the delta header. We replace both of these with an argument to __init__ containing a callable that can format a delta header given a revisiondelta instance. Differential Revision: https://phab.mercurial-scm.org/D4079

changegroup: make delta header struct formatters actual structs Why we weren't using compiled Struct instances, I don't know. They make code simpler. In theory they are faster. Although I don't believe I was able to measure any meaningful change. That could be because this code is often dominated by compression, deltafication, and function call overhead. Differential Revision: https://phab.mercurial-scm.org/D4078

changegroup: pass version into constructor Currently, the version is an attribute on each class. Passing the argument into the constructor gets us one step closer to eliminating cg2packer and cg3packer. Differential Revision: https://phab.mercurial-scm.org/D4077

changegroup: define functions for creating changegroup packers Currently, we have 3 classes for changegroup generation. Each class handles a specific changegroup format. And each subsequent version's class inherits from the previous one. The interface for the classes is not very well defined and a lot of version-specific behavior is behind overloaded functions. This approach adds complexity and makes changegroup generation difficult to reason about. Upcoming commits will be consolidating these 3 classes so differences between changegroup versions and changegroup generation are controlled by parameters to a single constructor / type rather than by overriding class attributes via inheritance. We begin this process by building dedicated functions for creating each changegroup packer instance. Currently they just call the constructor on the appropriate class. This will soon change. Differential Revision: https://phab.mercurial-scm.org/D4076

changegroup: capture revision delta in a data structure The current changegroup generation code is tightly coupled to the revlog API. This tight coupling makes it difficult to implement alternate storage backends without requiring a large surface area of the revlog API to be exposed. This is not desirable. In order to support changegroup generation with non-revlog storage, we'll need to abstract the concept of delta generation. This commit is the first step down that road. We introduce a data structure for representing a delta in a changegroup. The API still leaves a lot to be desired. But at least we now have separation between data and actions performed on it. As part of this, we tweak behavior slightly: we no longer concatenate the delta prefix with the metadata header. Instead, we track and emit the prefix as a separate chunk. This shouldn't have any meaningful impact since all the chunks just get sent to the wire, the compressor, etc. Because we're introducing a new object, this does add some overhead to changegroup execution. `hg perfchangegroupchangelog` on my clone of the Mercurial repo (~40,000 visible revisions in the changelog) slows down a bit: ! wall 1.268600 comb 1.270000 user 1.270000 sys 0.000000 (best of 8) ! wall 1.419479 comb 1.410000 user 1.410000 sys 0.000000 (best of 8) With for `hg bundle -t none-v2 -a /dev/null`: before: real 6.610 secs (user 6.460+0.000 sys 0.140+0.000) after: real 7.210 secs (user 7.060+0.000 sys 0.140+0.000) I plan to claw back this regression in future commits. And I may even do away with this data structure once the refactor is complete. For now, it makes things easier to comprehend. Differential Revision: https://phab.mercurial-scm.org/D4075

changegroup: inline ellipsisdata() There's only one caller of it. I don't think it needs to exist as a standalone function. Differential Revision: https://phab.mercurial-scm.org/D4074

changegroup: rename "revlog" variables "revlog" shadows the module import. But more importantly, changegroup generation should be storage agnostic and not assume the existence of revlogs. Let's rename the thing providing revision storage to "store" to reflect this ideal property. Differential Revision: https://phab.mercurial-scm.org/D4073

changegroup: move generate() modifications from narrow Narrow had a custom version of generate() that was essentially a copy of generate() with inline additions to facilitate ellipses serving. This commit inlines those modifications into generate(). Differential Revision: https://phab.mercurial-scm.org/D4067

changegroup: move generatefiles() from narrow The code is a bit ugly in that it overrides the linknodes function that is passed in as a function. I'd like to think that the caller of generatefiles() would pass in the appropriate function. We can clean this up later. Differential Revision: https://phab.mercurial-scm.org/D4066

changegroup: move _sortgroup() from narrow Differential Revision: https://phab.mercurial-scm.org/D4065

changegroup: move close() from narrow More of the same. Differential Revision: https://phab.mercurial-scm.org/D4064

changegroup: move revchunk() from narrow The monkeypatched revchunk for ellipses serving is a completely independent implementation. We model it as such in the changegroup code. revchunk() is now a simple proxy function. Again, I wish we had better APIs here. Especially since this narrow code is part of cg1packer and cg1packer can't be used with narrow. Class inheritance is wonky. And I will definitely be making changes to changegroup code for delta generation. As part of the code move, `node.nullrev` was replaced by `nullrev`. And a reference to `orig` was replaced to call `self._revchunknormal` directly. Differential Revision: https://phab.mercurial-scm.org/D4063

changegroup: move deltaparent() from narrow I'm not keen on performing the attribute sniff to test for presence of ellipses mode: I'd rather we use a separate packer instance that was ellipses mode specific. But I've tried to formalize a better API without narrow in core and I can't make sense of all the monkeypatching. My goal is to inline as much of the monkeypatching as possible then refactor the changegroup generation API. We add this code to the cg2packer because narrow doesn't work with cg1. Differential Revision: https://phab.mercurial-scm.org/D4062

changegroup: move _packellipsischangegroup() from narrow The behavior here is not ideal, as the function constructs a packer then adds attributes to it. This will be cleaned up in subsequent commits. Moving this code is necessary to move the remainder of the bundle2-level changegroup part generation code into core. Differential Revision: https://phab.mercurial-scm.org/D4061

changegroup: move ellipsisdata() from narrow This is a pretty straightforward copy of the function. Differential Revision: https://phab.mercurial-scm.org/D4060

fileset: narrow status computation by left-hand-side of 'and' node Timing with warm disk cache: $ hg status --cwd mozilla-central 'set:path:build/ and unknown()' --time (orig) time: real 1.970 secs (user 1.560+0.000 sys 0.410+0.000) (new) time: real 0.330 secs (user 0.310+0.000 sys 0.020+0.000)

fileset: move copy constructor of matchctx near __init__

fileset: build status according to 'withstatus' hint _switchcallers is no longer needed since 'withstatus' node is reinserted for arguments of functions like revs(). New matchctx instance is created per 'withstatus' to make sure that status tuple is available only for children of the 'withstatus' node.

fileset: insert hints where status should be computed This will allow us to compute status against a narrowed set of files. For example, "path:build/ & (unknown() + missing())" is rewritten as "path:build/ & <withstatus>(unknown() + missing(), 'unknown missing')", and the status call can be narrowed by the left-hand-side matcher, "path:build/". mctx.buildstatus() calls will be solely processed by getmatchwithstatus().

fileset: move buildstatus() to matchctx method In future patches, file status will be computed while evaluating a parsed tree. This patch provides a matchctx interface to build status.

fileset: keep basectx by matchctx

fileset: pass in basectx to _buildstatus() I'll make matchctx remember both ctx and basectx so that file status between them can be computed later. This prepares for the change.

resolve: update commands.resolve.confirm help text Included --mark and --unmark in the help text of resolve.confirm.config. Differential Revision: https://phab.mercurial-scm.org/D4103

resolve: support confirm config option with --unmark flag Now, commands.resolve.confirm also respect --unmark option; and confirm to unresolve all resolved files. It will confirm only when no files pats are passed (same as --mark), because when no pats are passed the default is to mark resolved files as unresolved. And if user has passed file pats then I think there is no need to confirm for that. Differential Revision: https://phab.mercurial-scm.org/D4102

resolve: correct behavior of mark-check=none to match docs Differential Revision: https://phab.mercurial-scm.org/D4121

narrow: move .hg/narrowspec to .hg/store/narrowspec (BC) The narrowspec is more closely related to the store than to the working copy. For example, if the narrowspec changes, the set of revlogs also needs to change (the working copy may change, but that depends on which commit is checked out). Also, when using the share extension, the narrowspec needs to be shared along with the store. This patch therefore moves the narrowspec into the store/ directory. This is clearly a breaking change, but I haven't bothered trying to fall back to reading the narrowspec from the old location (.hg/), because there are very few users of narrow out there. (We'll add a temporary hack to our Google-internal extension to handle the migration.) Differential Revision: https://phab.mercurial-scm.org/D4099

narrow: drop checkambig=True when restoring backup IIUC, checkambig is about updating timestamps of the file while renaming. That's important for the dirstate, but we never check the timestamp of the narrowspec file. We can therefore avoid checking passing checkambig=True. Differential Revision: https://phab.mercurial-scm.org/D4098

narrow: remove a repo file-cache invalidation It's unclear why this was needed. All tests pass without it. I asked Kyle Lippincott (who added the check) and he also doesn't remember what it was for. Differential Revision: https://phab.mercurial-scm.org/D4097

narrow: call narrowspec.{save,restore,clear}backup directly I want to move .hg/narrowspec to .hg/store/narrowspec and we need to decouple the narrowspec update from the dirstate update for that. This patch lets the callers call the narrowspec backup functions directly, in addition to the dirstate backup functions. The narrowspec methods are made to check if narrowing is enabled. For that, a repo instance was needed, which all the callers luckily already had available. Differential Revision: https://phab.mercurial-scm.org/D4096

index: don't add 1 to length variables A lot of "+ 1" and "-1" were mechanically added to ease the transition in 781b2720d2ac (index: don't include nullid in len(), 2018-07-20). Let's clean it up now. Differential Revision: https://phab.mercurial-scm.org/D4106

index: drop support for nullid at position len(index) in index_node I think no callers exist since at least a3dacabd476b (index: don't allow index[len(index)] to mean nullid, 2018-07-20). Differential Revision: https://phab.mercurial-scm.org/D4105

index: return False for "len(index) in index" Since we no longer accept index[len(index)], we should clearly make "len(index) in index" return False. This should have been part of a3dacabd476b (index: don't allow index[len(index)] to mean nullid, 2018-07-20) Differential Revision: https://phab.mercurial-scm.org/D4104

fileset: combine union of basic patterns into single matcher This appears to improve query performance in a big repository than I thought. Writing less Python in a hot loop, faster computation we gain. $ hg files --cwd mozilla-central --time 'set:a* + b* + c* + d* + e*' (orig) time: real 0.670 secs (user 0.640+0.000 sys 0.030+0.000) (new) time: real 0.210 secs (user 0.180+0.000 sys 0.020+0.000)

fileset: reorder 'or' expression by weight

fileset: introduce weight constants for readability These constants are defined in the filesetlang module since it's the bottommost module depending on WEIGHT_CHECK_FILENAME, and extensions will be likely to import it to process function arguments. Credit for the naming goes to Augie Fackler.