rust-dirstate-status: add `walk_explicit` implementation, use `Matcher` trait This is the first time we actually use the `Matcher` trait, still for a small subset of all matchers defined in Python. While I haven't yet actually measured the performance of this, I have tried to avoid any unnecessary allocations. This forces the use of heavy lifetimes annotations which I am not sure we can simplify, although I would be happy to be proven wrong. Differential Revision: https://phab.mercurial-scm.org/D7529

rust-matchers: add `FileMatcher` implementation Mercurial defines an `exactmatcher`, I find `FileMatcher` to be clearer, but am not opposed to using the old name. This change also switched the order of `assert_eq` arguments as it is clearer that way for most people. Differential Revision: https://phab.mercurial-scm.org/D7528

exchange: ensure all outgoing subrepo references are present before pushing We've run into occasional problems with people committing a repo, and then amending or rebasing in the subrepo. That makes it so that the revision in the parent can't be checked out, and the problem gets propagated on push. Mercurial already tries to defend against this sort of dangling reference by pushing *all* subrepo revisions first. This reuses the checks that trigger warnings in `hg verify` to bail on the push unless using `--force`. I thought about putting this on the server side, but at that point, all of the data has been transferred, only to bail out. Additionally, SCM Manager hosts subrepos in a location that isn't nested in the parent, so normal subrepo code would complain that the subrepo is missing when run on the server. Because the push command pushes subrepos before calling this exchange code, a subrepo will be pushed before the parent is verified. Not great, but no dangling references are exchanged, so it solves the problem. This code isn't in the loop that pushes the subrepos because: 1) the list of outgoing revisions is needed to limit the scope of the check 2) the loop only accesses the current revision, and therefore can miss subrepos that were dropped in previous commits 3) this code is called when pushing a subrepo, so the protection is recursive I'm not sure if there's a cheap check for the list of files in the outgoing bundle. If there is, that would provide a fast path to bypass this check for people not using subrepos (or if no subrepo changes were made). There's probably also room for verifying other references like tags. But since that doesn't break checkouts, it's much less of a problem. Differential Revision: https://phab.mercurial-scm.org/D7616

procutil: try and avoid angering CoreFoundation on macOS We've seen failures like this: objc[57662]: +[__NSCFConstantString initialize] may have been in progress in another thread when fork() was called. objc[57662]: +[__NSCFConstantString initialize] may have been in progress in another thread when fork() was called. We cannot safely call it or ignore it in the fork() child process. Crashing instead. Set a breakpoint on objc_initializeAfterForkError to debug. I think this is due to forking off some background processes during `hg update` or similar. I don't have any conclusive proof this is the fork() call that's to blame, but it's the most likely one since the regular `hg update` codepath uses the other fork() invocation (via workers) and we don't get this report from non-Google macOS users. Ugh. Differential Revision: https://phab.mercurial-scm.org/D7615

nodetree: simplify a conditionnal in shortesthexnodeidprefix instead of try to catch some attribute error, we could just nicely look if the attribute will be available. This make the code simpler to follow and less error prone since we no longer rely on a wider attribute catching. Differential Revision: https://phab.mercurial-scm.org/D7651

config: close file even if we fail to read it If we get an exception from cfg.read(), we would not close the file before this patch. This patch uses a context manager to make sure we close it. Differential Revision: https://phab.mercurial-scm.org/D7626

config: catch intended exception when failing to parse config When a new config parser was introduced in fca54469480e (ui: introduce new config parser, 2009-04-23), the reading side would raise a ConfigError which was then caught in the ui code. Then, in 2123aad24d56 (error: add new ParseError for various parsing errors, 2010-06-04), a ParseError was raised instead, but the call site was not updated. Let's start catching that ParseError. We still don't print it in a friendly way, but that's not worse than before. Differential Revision: https://phab.mercurial-scm.org/D7625

rust-hg-path: implement more readable custom Debug for HgPath{,Buf} The default prints the vector of bytes as a list of integers. I considered instead getting rid of the Debug trait, but we use the Debug format in lots of derived Debug instances, so we probably do want to implement it. Differential Revision: https://phab.mercurial-scm.org/D7604

util: implement sortdict.insert() As flagged by pytype (reported via Matt Harbison, thanks). This was broken by bd0fd3ff9916 (util: rewrite sortdict using Python 2.7's OrderedDict, 2017-05-16). We actually call insert() on namespaces.py:100, but we clearly don't have test coverage of that an no users have reported it AFAIK. Differential Revision: https://phab.mercurial-scm.org/D7680

patch: make __repr__() return str Caught by pytype: line 969, in __repr__: Function bytes.join was called with the wrong arguments [wrong-arg-types] Expected: (self, iterable: Iterable[bytes]) Actually passed: (self, iterable: Iterator[str]) Differential Revision: https://phab.mercurial-scm.org/D7682

pytype: suppress warnings about no 'open_binary' on importlib.resources Fixes these pytype warnings: line 43, in <module>: No attribute 'open_binary' on module 'importlib.resources' [module-attr] line 47, in open_resource: No attribute 'open_binary' on module 'importlib.resources' [module-attr] For some reason, I can't upgrade from 3.6.8 in my WSL environment. Differential Revision: https://phab.mercurial-scm.org/D7681

windows: if username(uid=None) is loaded, just use getpass This is at least consistent with what we do on other platforms in the base case. I don't know enough about Windows to fill in other cases that might exist here, but this at least should be a start. Differential Revision: https://phab.mercurial-scm.org/D7679

transplant: use check_incompatible_arguments() Differential Revision: https://phab.mercurial-scm.org/D7663

bookmarks: use check_incompatible_arguments() for inactive+action Differential Revision: https://phab.mercurial-scm.org/D7662

bookmarks: use cmdutil.check_incompatible_arguments() for action+rev Differential Revision: https://phab.mercurial-scm.org/D7648

bookmarks: use cmdutil.check_at_most_one_arg() for action Differential Revision: https://phab.mercurial-scm.org/D7647

rebase: use cmdutil.check_at_most_one_arg() for action Here we also needed to know what the action was (if any), so I've updated the helper to return any specified option. Differential Revision: https://phab.mercurial-scm.org/D7640

releasenotes: extract helper for checking for incompatible arguments This patch extracts a new check_incompatible_arguments() function similar to check_at_most_one_arg(). The difference is that the new function is for checking for arguments that are disallowed together with some other argument but not mutually exclusive among themselves. Differential Revision: https://phab.mercurial-scm.org/D7639

fix: use cmdutil.check_at_most_one_arg() Differential Revision: https://phab.mercurial-scm.org/D7638

patchbomb: use cmdutil.check_at_most_one_arg() Differential Revision: https://phab.mercurial-scm.org/D7637

export: use cmdutil.check_at_most_one_arg() Differential Revision: https://phab.mercurial-scm.org/D7636

amend: use cmdutil.check_at_most_one_arg() Differential Revision: https://phab.mercurial-scm.org/D7635

commit: use cmdutil.check_at_most_one_arg() Differential Revision: https://phab.mercurial-scm.org/D7634

clone: extract helper for checking mutually exclusive args We have some duplicated code for aborting if the user provided mutually exclusive arguments. Extensions surely have more such code. We also have duplicated translations and inconsistent output in this area. This patch introduces a simpler helper for checking if more than one option among a given set was given on the command line. I've made the clone code call the function to show that it works. The function has no good way of checking arguments with hyphens in them. I'll add that later if necessary. The function still won't be applicable in all cases, but I think it's still better than nothing. Differential Revision: https://phab.mercurial-scm.org/D7633

dirstate: when calling rebuild(), avoid some N^2 codepaths I had a user repo with 200k files in it. Calling `hg debugrebuilddirstate` took tens of minutes (I didn't wait for it). In that situation, changedfiles==allfiles, and both are lists. This meant that we had to run an average of 100k comparisons, for each of 200k files, just to check whether a file needed to have normallookup called (it always did), or drop. While it's probably not a huge issue, in my very awkward synthetic benchmark I wrote (not using a benchmark library or anything), I was seeing some slowdowns for small-changedfiles and very-large-allfiles invocations, with an inflection somewhere around 10 items in changedfiles (regardless of the size of allfiles); above 10 items in changedfiles, the new code appears to always be faster. For the case of 50k files in changedfiles and the same items in allfiles, I'm seeing differences of 15s of just running comparisons vs. 0.003793s. I haven't bothered to run a comparison of 200k items in changedfiles and allfiles. :) Differential Revision: https://phab.mercurial-scm.org/D7665

rust-warnings: fix warnings in tests It turns out that I also missed those warnings inside tests. This should be the last of them. One day we will get rid of this interface anyway. Differential Revision: https://phab.mercurial-scm.org/D7678

relnotes: mention the merging of index and nodemap

crecord: remove toggleamend Previous commit removed its only calling site.

crecord: repurpose "a" key to toggle all selections (BC) I really don't like "a". I keep accidentally hitting it when I actually want "A", and then I'm suddenly in a state I don't want to be in. There's a big wall of text telling me that I've turned amend mode on or off (which one was I orginally in?), and this seems very useless. If I wanted to amend or not, I would have chosen that from the command-line, not change my mind after I've already started picking hunks apart. Furthermore, for most uses of the hunk selector (revert, uncommit, shelve/unshelve), this amend toggle doesn't make sense. It seems much better to repurpose this key to be a "weaker" version of "A". It toggles all selections. This is pretty harmless if hit accidentally, (can just hit "a" again to toggle everything and undo it), and has immediate visual feedback that something happened: all the x's and blank spaces get switched around. And unlike with amend, the current flipped state is also immediately visible without having to read a wall of text. I'm calling this a BC, however, because somewhere, someone out there has probably really fallen in love with the old use of "a" and will get angry that we took it away.

hgweb: fix error in docstring Despite the subtle semantic difference, this sentence really meant to say "overridden", not "overwritten".