Sun, 11 Dec 2022 16:44:50 +0400 hghave: detect Python 3.10 and 3.11 as well stable
Anton Shestakov <av6@dwimlabs.net> [Sun, 11 Dec 2022 16:44:50 +0400] rev 49782
hghave: detect Python 3.10 and 3.11 as well Noticed because test-contrib-relnotes.t was skipped.
Mon, 05 Dec 2022 16:05:04 -0500 extensions: load help from hgext.__index__ as a fallback this time stable
Matt Harbison <matt_harbison@yahoo.com> [Mon, 05 Dec 2022 16:05:04 -0500] rev 49781
extensions: load help from hgext.__index__ as a fallback this time Prior to 843418dc0b1b, `hgext.__index__` was consulted first if present, which caused the longer help from the extension modules to be ignored, even when available. But that change causes a bunch of test failures when the pyoxidized binary bundles *.pyc in the binary, saying the there's no help topic for `hg help $disabled_extension` and suggesting the use of `--keyword`, rather than showing a summary and indicating that it is disabled. Current failures were in test-check-help.t, test-extension.t, test-help.t, and test-qrecord.t. Ideally, we would read the various *.pyc files from memory and slurp in the docstring, but I know that they used to not be readable as resources, and I can't figure out how to make it work now. So maybe 3.9 and/or the current PyOxidizer doesn't support it yet. I got closer in py2exe with `importlib.resources.open_binary("hgext", "rebase.pyc")`, but `open_binary()` on *.pyc fails in pyoxidizer.[1] Either way, the *.pyc can't be passed to `ast.parse()` as `extensions._disabledcmdtable()` is doing, so I'm setting that aside for now. [1] https://github.com/indygreg/PyOxidizer/issues/649
Wed, 07 Dec 2022 11:26:07 -0500 extensions: process disabled external paths when `hgext` package is in-memory stable
Matt Harbison <matt_harbison@yahoo.com> [Wed, 07 Dec 2022 11:26:07 -0500] rev 49780
extensions: process disabled external paths when `hgext` package is in-memory This fixes `hg help -e ambiguous` in test-helpt.t:2055 with the `ambiguous = !./ambiguous.py` configuration, when `hgext` is not in the filesystem (e.g. pyoxidizer builds with in-memory resources, or TortoiseHg with py2exe), but the disabled external extension is. Now instead of aborting with a suggestion to try `--keyword`, the help command prints text for the extension.
Mon, 05 Dec 2022 15:14:33 -0500 hg: show the correct message when cloning an LFS repo with extension disabled stable
Matt Harbison <matt_harbison@yahoo.com> [Mon, 05 Dec 2022 15:14:33 -0500] rev 49779
hg: show the correct message when cloning an LFS repo with extension disabled The `extensions._disabledpaths()` doesn't handle fetching help from `__index__`, so it returns an empty dictionary of paths. That means None is always returned from `extensions.disabled_help()` when embedding resources inside the pyoxidizer or py2exe binary, regardless of the arg or if is an external extension stored in the filesystem. And that means wrongly telling the user with an explicitly disabled LFS extension that it will be enabled locally upon cloning from an LFS remote. That causes test-lfs-serve.t:295 to fail. This effectively reverts most of the rest of 843418dc0b1b, while keeping the help text change in place (which was specifically identified as a problem).
(0) -30000 -10000 -3000 -1000 -300 -100 -30 -10 -4 +4 +10 +30 +100 +300 +1000 tip