transaction: actually delete file created during the transaction on rollback Transaction currently has two modes: - one where file created during the transaction are deleted on rollback, - one where file created during the transaction are truncated to 0 on rollback. Before this change, `hg rollback` and `hg recover` are using the "delete" mode and transaction abort is using the "truncate" option. This difference is never really explained. A long time ago, there was two code paths, with this divergence existing for unclear reasons. When the two code paths got merged into a single one, a boolean argument have been added to preserve this divergence, mostly probably as a cargo cult. The divergence is weird and induce bad surprises, and the truncate behavior is a bit odds, introducing other bad surprises (e.g. 08ecbdba186f) So solve this, we stop using the "truncate" behavior and unify on the "delete" behavior. Despite being currently more "common", the truncate behavior seems less natural, resulting in the transaction leaving empty file around. This is landed on default, early in the cycle, to help us catch problems that could emerge.

rhg: tweak rhg fallback code structure The goal here is to share the code that accesses the config setting, but also move it to the top-level, so that we can more easily change this code later. (in fact in Jane Street we want to make it choose chg in some cases)

journal: track bookmark deletion Bookmark deletion are now properly tracked by the journal extensions.

journal: show that we do not track deletion This was an oversight for the initial implementation.

contrib: upgrade merge-lists from clap v3 to clap v4 Mostly to avoid having clap v3 in our dependencies, due to https://osv.dev/vulnerability/GHSA-g98v-hv3f-hcfr (even though that seems pretty harmless).

compat: use cp -R in `test-revlog-delta-find.t` (issue6835) This follow POSIX and make things compatible with NetBSD.

compat: use cp -R in `test-bundle-phase-internal.t` (issue6835) This follow POSIX and make things compatible with NetBSD.

infinitepush: drop the extension The writing have been on the wall for a long time.

perf: display all timing by default This is much more useful and end up enable it everywhere. I don't think we have strong backward compatibility guarantee for perf.

wrapfunction: deprecated calling "wrapfunction" with bytes This will help us to detect and eradicates legacy calls. This is on the path to deprecating `util.safehasattr`.

wrapfunction: deprecates calling `wrappedfunction` with bytes This will help us to detect and eradicates legacy calls. This is on the path to deprecating `util.safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in various tests This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "exthelper" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "sqlitestore" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "share" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "perf" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "zeroconf" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "sqlitestore" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "sparse" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "schemes" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "remotenames" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "remotefilelog" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "pager" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "narrow" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "lfs" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "largefiles" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "keyword" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "journal" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "infinitepush" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "histedit" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "highlight" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in the "git" extension This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "fsmonitor" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "fastannotate" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "clonebundles" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "bookflow" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

wrapfunction: use sysstr instead of bytes as argument in "beautifygraph" This is as valid and simpler, it will help us to eventually get ride of `safehasattr`.

byteify-strings: passe sysstr to attr function and wrapper These function now accepts string so we should not do byteify them anymore.

repoview: fix the filter created by `extrafilter` The `name` variable was updated in each interaction of the loop, making all new filters inheriting from the same base (the last one iterated), leading to terrible terrible misbehavior.

repoview: show `extrafilter` misbehave when other filtering is in play See new changeset for fix and details.

rust-config: add docstring to inner `get_parse` method The signature is quite heavy, a docstring to explain the idea doesn't hurt.

rust-config: add devel warning when using undeclared config items This mirrors the Python implementation now that we're done catching up.

rhg: use default configitem in `cat` Now that we've declared the items, we can get rid of the manual fallback.

rhg: add config defaults to configitems.toml This will allow us to use the same default config mechanic as the rest of the code.

rust-blackbox: use `is_extension_enabled` config helper It's there, may as well use it to make the code clearer and less bug-prone.

configitems: move blackbox's config items to the new configitems.toml In order for the Rust code to gain access to default values of in-core extensions that have a Rust implementation, we need to centralize them alongside the core items declarations. This is the first and so far only one of the extensions that have gained Rust support, I don't think it's worth the churn to move the rest of the extension's configitems yet.

rust-config: add config getters that don't fall back to defaults This is useful in cases where we access config items that are more... lenient with their types than a fresh new system would allow. For now there is only a single use of this, but we might get more later.

rust-config: add support for default config items Now that configitems.toml exists, we can read from it the default values for all core config items. We will add the devel-warning for use of undeclared config items in a later patch when we're done adding the missing entries for `rhg`.

configitems: declare items in a TOML file Mercurial ships with Rust code that also needs to read from the config. Having a way of presenting `configitems` to both Python and Rust is needed to prevent duplication, drift, and have the appropriate devel warnings. Abstracting away from Python means choosing a config format. No single format is perfect, and I have yet to come across a developer that doesn't hate all of them in some way. Since we have a strict no-dependencies policy for Mercurial, we either need to use whatever comes with Python, vendor a library, or implement a custom format ourselves. Python stdlib means using JSON, which doesn't support comments and isn't great for humans, or `configparser` which is an obscure, untyped format that nobody uses and doesn't have a commonplace Rust parser. Implementing a custom format is error-prone, tedious and subject to the same issues as picking an existing format. Vendoring opens us to the vast array of common config formats. The ones being picked for most modern software are YAML and TOML. YAML is older and common in the Python community, but TOML is much simpler and less error-prone. I would much rather be responsible for the <1000 lines of `tomli`, on top of TOML being the choice of the Rust community, with robust crates for reading it. The structure of `configitems.toml` is explained inline.

thirdparty: vendor tomli The next commit will introduce a .toml file to abstract configitems away from Python. Python 3.11 has a toml read-only library (`tomllib`), which gives us a way out of vendoring eventually. For now, we vendor the backport, specifically version 1.2.3 which is still compatible with Python 3.6.

configitems: add `documentation` field It may be useful to expose documentation information in the help in some form. This will be populated in a future changeset by using the current comments that are relevant for users.

configitems: use standard "dynamicdefault" approach in edge case This makes for fewer edge cases, which will help a future patch.

configitems: fix typo in devel warning about extension overrides

extensions: imp module is removed in Python 3.12 - use importlib to load files imp has been deprecated for a long time, and has finally been removed in Python 3.12 . imp was only used for loading extensions that has been specified with direct .py path or path to a package directory. The same use cases can be achieved quite simple with importlib, , possiby with small changes in corner cases with undefined behaviour, such as extensions without .py source. There might also be corner cases and undefined behaviour around use of sys.modules and reloading.

utils: imp module is removed in Python 3.12 - get is_frozen() from _imp imp has been deprecated for a long time, and has finally been removed in Python 3.12 . The successor importlib is using the same internal _imp module as imp, but doesn't expose it's is_frozen. Using the internal function directly seems like the cleanest solution. Another alternative to imp.is_frozen("__main__") is sys.modules['__main__'].__spec__.origin == 'frozen' but that seems even more internal and fragile.

extensions: address ast deprecations introduced in Python 3.12 Tests would fail with: .../mercurial/extensions.py:910: DeprecationWarning: ast.Str is deprecated and will be removed in Python 3.14; use ast.Constant instead if isinstance(a, ast.Str): .../mercurial/extensions.py:912: DeprecationWarning: ast.Bytes is deprecated and will be removed in Python 3.14; use ast.Constant instead elif isinstance(a, ast.Bytes): .../mercurial/extensions.py:913: DeprecationWarning: Attribute s is deprecated and will be removed in Python 3.14; use value instead name = a.s

vfs: handle shutil.rmtree deprecation of onerror in Python 3.12 Tests would fail with warnings: .../mercurial/vfs.py:289: DeprecationWarning: onerror argument is deprecated, use onexc instead The excinfo changed slightly, but we don't use it anyway.

tests: fix sortdict doctest with Python 3.12 The output of OrderedDict changed to use plain dict syntax: $ python3.11 -c "import collections;print(collections.OrderedDict([('a', 0), ('b', 1)]))" OrderedDict([('a', 0), ('b', 1)]) $ python3.12 -c "import collections;print(collections.OrderedDict([('a', 0), ('b', 1)]))" OrderedDict({'a': 0, 'b': 1})

utils: stop using datetime.utcfromtimestamp() deprecated in Python 3.12 Python3.12 made tests fail with warnings: DeprecationWarning: datetime.utcfromtimestamp() is deprecated and scheduled for removal in a future version. Use timezone-aware objects to represent datetimes in UTC: datetime.fromtimestamp(timestamp, datetime.UTC). Computing the diff while in timestamp seconds seems to preserve to the original intent from ae04af1ce78d. It would be nice to have some doctest coverage of this, with the problematic corner cases that has popped up over time...

branching: merge stable into default