requirements: move loading to hg-core and add parsing No functional change, checking comes later. Differential Revision: https://phab.mercurial-scm.org/D9398

rhg: add a `debugrequirements` subcommand For now it only prints the contents of `.hg/requires` as-is, without parsing. Differential Revision: https://phab.mercurial-scm.org/D9397

pyoxidizer: make sure defaultrc directory exists before trying to write to it When doing some work involving one-file binaries, this line is failing for me. It seems reasonable to just make sure the destination directory exists before splatting the file into it. Differential Revision: https://phab.mercurial-scm.org/D9401

commands: fix checking of share safe requirement on `config --shared` The `if requirements.SHARESAFE_REQUIREMENT in ...` was wrongly placed inside another if statement which made the check unreachable. Differential Revision: https://phab.mercurial-scm.org/D9360

dispatch: pass root path in ui.readconfig() as root of main repo Since we are reading main (shared-source) repository config options, we should pass root as that repository root only. Differential Revision: https://phab.mercurial-scm.org/D9359

scmutil: try-delete `.hg/store/requires` if store requirements are empty When downgrading from a shared-safe repository to non-shared-safe repository, we end up in a case where we had requirements stored in `.hg/store/requires` but no longer want them there. Let's explicitly try delete the `.hg/store/requires` file if store requirements are empty. Differential Revision: https://phab.mercurial-scm.org/D9357

errors: raise InputError on bad top-level flags Differential Revision: https://phab.mercurial-scm.org/D9388

errors: raise StateError on uncommitted changes when merge starts Differential Revision: https://phab.mercurial-scm.org/D9393

errors: raise StateError when there are unresolves merge conflicts Differential Revision: https://phab.mercurial-scm.org/D9392

errors: introduce SecurityError and use it in a few places This is part of https://www.mercurial-scm.org/wiki/ErrorCategoriesPlan. There are perhaps more errors in `sslutil.py` that should raise `SecurityError`; I picked the most clear ones to start with. Differential Revision: https://phab.mercurial-scm.org/D9390

errors: raise InputError when too few arguments given to alias Differential Revision: https://phab.mercurial-scm.org/D9389

errors: raise InputError on bad bookmark argument Differential Revision: https://phab.mercurial-scm.org/D9385

errors: raise ConfigError on bad alias definition Differential Revision: https://phab.mercurial-scm.org/D9384

errors: raise InputError on bad repo arguments I'm not sure if one of these should be StateError... Differential Revision: https://phab.mercurial-scm.org/D9383

errors: drop trailing "!" for some errors about bookmarks Differential Revision: https://phab.mercurial-scm.org/D9382

errors: remove trailing "!" in messages about bad top-level args Differential Revision: https://phab.mercurial-scm.org/D9381

errors: remove trailing "!" in messages about creating new heads on push Differential Revision: https://phab.mercurial-scm.org/D9380

errors: consistently don't use trailing "!" in "not found in manifest" message We already had some places with the same message without the trailing "!". Differential Revision: https://phab.mercurial-scm.org/D9379

errors: remove trailing "!" from some error messages for consistency Some types of exceptions had a trailing "!" printed after the message from the exception itself. I guess some of these errors seem a little more severe (?), but it seems more likely that the inconsistency was just an oversight. Differential Revision: https://phab.mercurial-scm.org/D9378

bisect: use tuple literal instead of split on string literal Differential Revision: https://phab.mercurial-scm.org/D9371

hgignore: add VS Code config Differential Revision: https://phab.mercurial-scm.org/D9370

tests: make test-worker.t pass on py2 I broke the py2 version in https://phab.mercurial-scm.org/D9287 because the `WorkerError.__bytes__()` (or `.__str__()`?) output was different in py2 compared to py3. Part of the problem was that I didn't propagate the status code that was passed in to the superclass so it could get printed. This patch fixes that. I don't know how it worked on py3 before this patch... I also added the usual `__bytes__ = _tobytes` override for good measure. It doesn't seem to be needed for tests to pass, though. Differential Revision: https://phab.mercurial-scm.org/D9377

tests: update test-https.t's #no-defaultcacertsloaded block with new exit code I'm pretty sure I broke this in https://phab.mercurial-scm.org/D9309. It was reported by the Heptapod CI. Differential Revision: https://phab.mercurial-scm.org/D9376

chg: fix test-check-clang-format.t failure Differential Revision: https://phab.mercurial-scm.org/D9365