windows: disable pager when packaged with py2exe With Windows and py3, all output that got directed to the pager was lost. It can be worked around by the user piping to `more`, but that's easy to forget, and can be dangerous if `hg diff` or similar incorrectly shows no changes. The problem appears to be the new WindowsConsoleIO in py3.6[1]. We've worked around it with PyOxidizer by setting the `Py_LegacyWindowsStdioFlag` interpreter option, and worked around it with `hg.bat` and `exewrapper.c` by internally setting `PYTHONLEGACYWINDOWSSTDIO=1`. Unfortunately, py2exe doesn't appear to be able to set the interpreter option, and somehow seems to also ignore the environment variable. The latter isn't a good fix anyway, since setting it in the environment would affect other python programs too. We can't install a global config for this because a config closer to the user (e.g. from before pager was turned on by default) can override it. [1] https://peps.python.org/pep-0528/ Differential Revision: https://phab.mercurial-scm.org/D12556

resourceutil: force filesystem access to resources when using py2exe I don't know why it doesn't work, but it avoids this fatal error on startup: > hg debugshell Traceback (most recent call last): File "hg", line 58, in <module> File "mercurial\dispatch.pyc", line 143, in run File "mercurial\dispatch.pyc", line 232, in dispatch File "mercurial\dispatch.pyc", line 254, in _rundispatch File "mercurial\ui.pyc", line 316, in load File "mercurial\rcutil.pyc", line 98, in rccomponents File "mercurial\rcutil.pyc", line 68, in default_rc_resources File "mercurial\utils\resourceutil.pyc", line 102, in contents File "<frozen zipimport>", line 775, in contents AssertionError I assume the py2 version of py2exe never hit this because `importlib.resources` failed to import. Differential Revision: https://phab.mercurial-scm.org/D12554

setup: fix the py2exe logic to work with py3 TortoiseHg still uses (the modernized) py2exe packaging, but the build was failing since `py2exe.Distribution` was removed. One thing to note is that later in this module, there's a hack to include `distutils` when building from a virtualenv. While `import distutils` works in `hg debugshell` when built with py2, it doesn't work in py3. I'm not sure why- I don't see it in `library.zip` either. It doesn't seem to break anything though. Differential Revision: https://phab.mercurial-scm.org/D12553

absorb: make `--edit-lines` imply `--apply-changes` One of our users tried to use `hg absorb -e` but it seemed that it would only bring up the editor if there were no changes the command could automatically detect destination for. I spent probably half an hour debugging why it worked that way. I finally figured out that it does bring up the editor, but you have to answer "yes" to the "apply changes" prompt *first*. That seems very unintuitive. If the user wants to edit the changes, there seems to be little reason to present them with a prompt first, so let's have `-e/--edit-lines` imply `-a/--apply-changes`. All the tests using `-e` also already used `-a`. I changed them to rely on the implied `-a` so we get coverage of that. Differential Revision: https://phab.mercurial-scm.org/D12550

branchmap: add a test that shows bad interaction with strip Differential Revision: https://phab.mercurial-scm.org/D12549

rhg: refactor to pass argv down, instead of caling args_os() This refactoring makes it easy to patch some command-line preprocessing into rhg. We use this to support using rhg as a shebang interpreter, for example. Differential Revision: https://phab.mercurial-scm.org/D12543

narrow: support debugupgraderepo Differential Revision: https://phab.mercurial-scm.org/D12542

procutil: avoid `+= None` when writing to full std{err,out} descriptor on py3 The write function returns `None` if there was no room to write the given data[1]. I don't like that this is effectively an infinite loop if there's never any progress emptying the underlying buffer, but we're no worse off than before, and it fixes random stacktrace popups seen in the py3 build of TortoiseHg. [1] https://docs.python.org/3/library/io.html#io.RawIOBase.write Differential Revision: https://phab.mercurial-scm.org/D12555

amend: don't remove unselected removals from memctx When there are removed files in the working copy and they are not selected to be amended into the parent, the `filectxfn` we create for the `memctx` would still return `None` before this patch. That's clearly incorrect; we should return the `filectx` from the unamended commit. Somehow it seems to not matter much except for the case with copies stored in changesets. Thanks to Kyle Lippincott for doing all the debugging and identifying the fix for this issue. Differential Revision: https://phab.mercurial-scm.org/D12573

tests: demonstrate crash on partial amend with copies in changesets See the fix in the next patch for explanation. Differential Revision: https://phab.mercurial-scm.org/D12572

rust-revlog: add methods for getting parent revs and entries Differential Revision: https://phab.mercurial-scm.org/D12442

rust-changelog: start parsing changeset data This patch makes `ChangelogRevisionData` do some coarse, line-level splitting of the changeset data into manifest node, user, timestamp, files list, and description. There are no (in-tree) users of these functions yet, but I've added tests to prevent regressions. We'll surely add callers at some point. Differential Revision: https://phab.mercurial-scm.org/D12439

rust-changelog: remove special parsing of empty changelog data for null rev For the null revision, `Revlog::get_rev_data()` will return an empty string (of bytes). We currently handle that case in `ChangelogRevisionData::manifest_node()`. However, it's going to be ugly to have special handling for the null revision for each future method on `ChangelogRevisionData`. This patch therefore restructures the code so we instead initialize the struct with valid data for the null revision. Differential Revision: https://phab.mercurial-scm.org/D12438

rust-changelog: don't skip empty lines when iterating over changeset lines The first empty line in the changeset indicates the end of headers and beginning of description. Callers can't know figure out where that position is if empty lines are skipped. Differential Revision: https://phab.mercurial-scm.org/D12426