persistent-nodemap: introduce a test to highlight possible race Weakness in the current file caching of the changelog means that a writer can end up using an outdated docket. This might result in "committed" persistent-nodemap data from a previous writer to be overwritten by a later writer. This break the strong "append only" assumption of the persistent nodemap and can result in confused reader. The race windows are quite narrow. See the test documentation for details. The issues is fixed in the next changeset. Differential Revision: https://phab.mercurial-scm.org/D11481

test: enable share-safe in test-persistent-nodemap We will use some share in this test and I would rather have them safe. Differential Revision: https://phab.mercurial-scm.org/D11480

filecache: abstract the fetching of the list of tracked file We will need it for a coming fix that will requires to check a variable list of file for the changelog. Differential Revision: https://phab.mercurial-scm.org/D11479

persistent-nodemap: use quiet upgrade in tests This make them less verbose while keeping the important information in the checked output. Differential Revision: https://phab.mercurial-scm.org/D11478

persistent-nodemap: fix a typo in a test comment oops. Differential Revision: https://phab.mercurial-scm.org/D11477

bookmarks: add an option to make pull mirror remote bookmarks For backups for instance. Merging bookmarks is not a useful behavior in that case. Differential Revision: https://phab.mercurial-scm.org/D11490

errors: use InputError for bad path arguments to `hg annotate` Differential Revision: https://phab.mercurial-scm.org/D11498

errors: use InputError for bad --similarity value Differential Revision: https://phab.mercurial-scm.org/D11497

errors: use InputError for some invalid revsets and such Differential Revision: https://phab.mercurial-scm.org/D11496

merge: with stable

dirstate: fix leak of entry object in dirstate_item_from_v1_data()

dirstate: fix parse_dirstate() to error out if NULL entry created Since 83f0e93ec34b "dirstate-item: move the C implementation to the same logic", dirstate_item_from_v1_data() gets more likely to return NULL, and the fuzzer crashes because of that.

rust: Remove EntryState::Unknown This enum variant represented the `state == '?'` case, which was used to represent the absence of a dirstate entry/item (and therefore of that entry’s state). Now that previous refactors have removed this use in the Python/Rust FFI APIs, the remaining uses can be removed by replacing `EntryState` by `Option<EntryState>` where appropriate, using `None` to represent the absence of an entry. Differential Revision: https://phab.mercurial-scm.org/D11465

dirstate: Remove the `state == ' '` special case Previously this was used to create a `DirstateItem` representing dirstate tree nodes that semantically don’t have an associated `DirtateItem`. This isn’t used anymore now that `dirstatemap.debug_iter` yields plain tuples. Differential Revision: https://phab.mercurial-scm.org/D11464

debugsate: Change debug_iter() to yield tuples instead of DirstateItem This removes the need for `DirstateItem` to support the `state == ' '` special case which represents dirstate tree nodes without an item. Differential Revision: https://phab.mercurial-scm.org/D11463

debugstate: Always call dirstatemap.debug_iter() … passing it a new `all` argument for the `--all` CLI option, instead of conditionally calling `.debug_iter()` or `.items()` This prepares for the next commit. Differential Revision: https://phab.mercurial-scm.org/D11462