update: add a Rust fast-path when updating from null (and clean)
This case is easy to detect and we have all we need to generate a valid
working copy and dirstate entirely in Rust, which speeds things up
considerably:
On my machine updating a repo of ~300k files goes from 10.00s down to 4.2s,
all while consuming 50% less system time, with all caches hot.
Something to note is that further improvements will probably happen
with the upcoming `InnerRevlog` series that does smarter
mmap hanlding, especially for filelogs.
Here are benchmark numbers on a machine with only 4 cores (and no SMT enabled)
```
### data-env-vars.name = heptapod-public-2024-03-25-ds2-pnm
# benchmark.name = hg.command.update
# bin-env-vars.hg.py-re2-module = default
# bin-env-vars.hg.changeset.node = <this change>
# benchmark.variants.atomic-update = no
# benchmark.variants.scenario = null-to-tip
# benchmark.variants.worker = default
default: 5.328762 ~~~~~
rust: 1.308654 (-75.44%, -4.02)
### data-env-vars.name = mercurial-devel-2024-03-22-ds2-pnm
# benchmark.name = hg.command.update
# bin-env-vars.hg.py-re2-module = default
# bin-env-vars.hg.changeset.node = <this change>
# benchmark.variants.atomic-update = no
# benchmark.variants.scenario = null-to-tip
# benchmark.variants.worker = default
default: 1.693271 ~~~~~
rust: 1.151053 (-32.02%, -0.54)
### data-env-vars.name = mozilla-unified-2024-03-22-ds2-pnm
# benchmark.name = hg.command.update
# bin-env-vars.hg.py-re2-module = default
# bin-env-vars.hg.changeset.node = <this change>
# benchmark.variants.atomic-update = no
# benchmark.variants.scenario = null-to-tip
# benchmark.variants.worker = default
default: 38.901613 ~~~~~
rust: 11.637880 (-70.08%, -27.26)
### data-env-vars.name = netbsd-xsrc-public-2024-09-19-ds2-pnm
# benchmark.name = hg.command.update
# bin-env-vars.hg.py-re2-module = default
# bin-env-vars.hg.changeset.node = <this change>
# benchmark.variants.atomic-update = no
# benchmark.variants.scenario = null-to-tip
# benchmark.variants.worker = default
default: 4.793727 ~~~~~
rust: 1.505905 (-68.59%, -3.29)
```
= Mercurial 6.0.3 =
* Fix an infinite loop in edge cases of the pure Python dirstate-v2 packer
* Fix a small race condition with deleted files in Rust-augmented status
* Small improvement to Python 3.11 compatibility
* Fixed some typos in help messages that caused missing sections
* Centos -> Rockylinux packaging
= Mercurial 6.0.2 =
* Fix `hg incoming` and `hg outgoing` with remote subrepos on Windows
* Fix Rust build on MacOS
* Fix a (15 year old?) bug where some data loss could happen in a very tight race window (f38ae2d7390e + ccd9cb73125c)
* Remove support for running `hg unamend` on merge changesets at it was never actually implemented
* Fix stream-clone requirements filtering which caused some stream-clones to be unusable from an older version of Mercurial
* Rename dirstate-v2 configuration name from `exp-rc-dirstate-v2` to `use-dirstate-v2`. This was overlooked in `6.0`, the old name will be kept as an alias for compatibility.
* Fix a bug where the branchmap could point to uncommitted data
= Mercurial 6.0.1 =
* Improve documentation around Rust (see hg help rust) and rust-related actions
* Improve upgrade/downgrade edge cases for dirstate-v2
* Make the test suite nicer to big-endian platforms
* Make the test suite nicer to NetBSD
* Fix a performance regression on fsmonitor (issue6612)
* Add fixes for Python 3.10
* Fix a deadlock when using sparse and share-safe together
= Mercurial 6.0 =
== New Features ==
* `debugrebuildfncache` now has an option to rebuild only the index files
* a new `bookmarks.mode` path option have been introduced to control the
bookmark update strategy during exchange with a peer. See `hg help paths` for
details.
* a new `bookmarks.mirror` option has been introduced. See `hg help bookmarks`
for details.
* more commands support detailed exit codes when config `ui.detailed-exit-codes` is enabled
== Default Format Change ==
== New Experimental Features ==
* '''Major feature''': version 2 of the dirstate is available (the first version is as old as Mercurial itself). It allows for much faster working copy inspection (status, diff, commit, update, etc.) and richer information (symlink and exec info on Windows, etc.). The format has been frozen with room for some future evolution and the current implementations (Python, Python + C, Python + Rust or pure Rust) should be compatible with any future change or optimization that the format allows. You can get more information [[https://www.mercurial-scm.org/repo/hg/file/tip/mercurial/helptext/internals/dirstate-v2.txt | in the internal documentation]]
* Added a new `web.full-garbage-collection-rate` to control performance. See
de2e04fe4897a554b9ef433167f11ea4feb2e09c for more information
* Added a new `histedit.later-commits-first` option to affect the ordering of commits in `chistedit` to match the order in `hg log -G`. It will affect the text-based version before graduating from experimental.
== Bug Fixes ==
* `hg fix --working-dir` now correctly works when in an uncommitted merge state
* Unintentional duplicated calls to `hg fix`'s internals were removed, making it potentially much faster
* `rhg cat` can be called without a revision
* `rhg cat` can be called with the `.` revision
* `rhg cat` is more robust than before with regards to edge cases. Some still remain like a tag or bookmark that is ambiguous with a nodeid prefix, only nodeids (prefixed or not) are supported as of now.
* `rhg cat` is even faster
* `rhg` (Rust fast-path for `hg`) now supports the full config list syntax
* `rhg` now parses some corner-cases for revsets correctly
* Fixed an `fsmonitor` on Python 3 during exception handling
* Lots of Windows fixes
* Lots of miscellaneous other fixes
* Removed a CPython-specific compatibility hack to improve support for alternative Python implementations
== Backwards Compatibility Changes ==
== Internal API Changes ==
The following functions have been removed:
* `dirstate.normal`
* `dirstate.normallookup`
* `dirstate.otherparent`
* `dirstate.add`
* `dirstate.addfile`
* `dirstate.remove`
* `dirstate.drop`
* `dirstate.dropfile`
* `dirstate.__getitem__`
* `dirstatemap.nonnormalentries`
* `dirstatemap.nonnormalset`
* `dirstatemap.otherparentset`
* `dirstatemap.non_normal_or_other_parent_paths`
* `dirstateitem.dm_nonnormal`
* `dirstateitem.dm_otherparent`
* `dirstateitem.merged_removed`
* `dirstateitem.from_p2`
* `dirstateitem.merged`
* `dirstateitem.new_merged`
* `dirstateitem.new_added`
* `dirstateitem.new_from_p2`
* `dirstateitem.new_possibly_dirty`
* `dirstateitem.new_normal`
* `dirstateitem.from_p2_removed`
Miscellaneous:
* `wireprotov1peer`'s `batchable` is now a simple function and not a generator
anymore
* The Rust extensions (and by extension the experimental `rhg status`) only use a tree-based dirstate in-memory, even when using dirstate-v1. See bf8837e3d7cec40fe649c47163a3154dda03fa16 for more details
* The Rust minimum supported version is now 1.48.0 in accordance with out policy of keeping up with Debian stable
* The test harness plays nicer with the NixOS sandbox