transaction: fix __repr__() and make the default name bytes This likely was always wrong on py3, since going back to aff5996f3043, these were added as a r-strings. Callers seem to always be supplying bytes, which makes the `b'/'.join(...)` part OK, but then bytes can't be interpolated into str with "%s", so it wouldn't have worked in either case.

rhg: support rhg files [FILE] This comes mostly for free after the rhg status [FILE] implementation.

rhg: support "status FILE" This change adds a new [file] argument to rhg status, parses them as patterns, canonicalizes the paths, and constructs a new PatternMatcher to intersect with the existing matcher being passed to the status implementation. We also make filepatterns a public module so we can access the pattern-parsing functionality we need from commands/status.rs.

rust-matchers: add PatternMatcher This should match the behavior of the Python patternmatcher. We need this for status [FILES] support.

rust-status: only visit parts of the tree requested by the matcher This is an optimization that the matcher is designed to support, but we weren't doing it until now. This is primarily relevant for supporting "hg status [FILES]", where this optimization is crucial for getting good performance (without this optimization, that command will still scan the entire tree, and just filter it down after the fact). When this optimization fires we have to return false from traverse_fs_directory_and_dirstate, representing that that part of the tree *might* have new files which we didn't see because we skipped parts of it. This only affects the cached result of the status, and is necessary to make future status operations (which might use a different matcher) work properly.

rust-status: error on non-existent files in file_set file_set here consists of the files that were explicitly specified on the command line. Erroring on them matches the behavior of Python status.

rust-status: explicitly track bad file types Before this, we silently skipped bad file types. Now, we check to see if a path is an exact_match in our matcher, and if so, print an error for it, and only then do we skip it. A path will be an exact match when it's specified as an explicit command line argument, and this error-printing behavior is necessary for compatibility with Python status.

rust: de-hardcode glob_suffix We're adding patternmatcher in a subsequent commit, and this needs needs to be different for includematcher and patternmatcher.

rust: improve the type on DirsMultiset::from_manifest It could only return an HgPathError, but we didn't express this in the type, so we needed some unreachable!()s. Now that is expressed in the type.

rust: don't escape spaces in regex Spaces are not in fact a regex special character, and escaping them is not correct.

rust-filepatterns: also normalize RelPath These patterns should be normalized too; this at least does no harm, and is necessary for proper status support.

rust: simplify pattern file parsing Previously we kept the pattern syntax as a &[u8] until the last possible moment, which meant we had to handle potential errors when parsing that &[u8]. But such errors could never actually occur, given the structure of the code. Now we parse it eagerly (in two places) and pass it around as a PatternSyntax, so we can delete some error handling code. parse_one_pattern is in particular useful for parsing patterns passed on the command line, as we'll support later in this series.

rust-matchers: better support file_set in IntersectionMatcher Previously, if both of the input matchers had non-exact file sets, IntersectionMatcher would simply have an empty fileset. Now, we duplicate Python behavior: we *union* the filesets of the matchers in that case. This makes some sense, because without exact input file sets, we can't do an exact intersection and must over-approximate. Concretely, this is necessary because the file_set is the list of files that were explicitly listed by the user, and we want to provide explicit errors for all such files, from both matchers.

branching: merge stable into default

version: migrate `opts` to native kwargs

verify: migrate `opts` to native kwargs

tags: migrate `opts` to native kwargs

tag: migrate `opts` to native kwargs

summary: migrate `opts` to native kwargs This has the added benefit of unconfusing PyCharm- it thought `opts` in the nested functions were str based for some reason.

revert: migrate `opts` to native kwargs

rename: drop an unnecessary variable declaration at a wider scope

rename: migrate `opts` to native kwargs

remove: migrate `opts` to native kwargs

purge: migrate `opts` to native kwargs

pull: avoid referencing a variable before it is initialized If `urlutil.get_pull_paths()` somehow yields nothing, the reference after the loop appears to be a problem. I noticed this because PyCharm flagged (and still flags) the initialization of this inside the loop as somehow unused. I'm not sure why 1 isn't returned directly where it is set to the result of `postincoming()`, but maybe the code is structured this way in anticipation of future work.

pull: migrate `opts` to native kwargs

paths: migrate `opts` to native kwargs

merge: migrate `opts` to native kwargs

manifest: migrate `opts` to native kwargs

locate: migrate `opts` to native kwargs

import: migrate `opts` to native kwargs

heads: migrate `opts` to native kwargs

forget: migrate `opts` to native kwargs

copy: drop an unused variable

copy: migrate `opts` to native kwargs

config: use the writefile() utility to create the sample hgrc This function uses a context manager.

config: migrate `opts` to native kwargs

clone: migrate `opts` to native kwargs

bundle: migrate `opts` to native kwargs

branches: migrate `opts` to native kwargs

branch: migrate `opts` to native kwargs

bookmark: migrate `opts` to native kwargs There are a bunch of callers for `ui.formatter()`, so convert to bytes only for that call.

archive: migrate `opts` to native kwargs I'm not sure how to handle migrating core APIs like `scmutil.match`, so kick that can down the road. (Plus it's used all over, so the callers need to be migrated first.)

backout: migrate `opts` to native kwargs It will take a bit to unwind `cmdutil.commit`, so there's a conversion to byteskwargs there, without changing the type of `opts` in this function. That's also useful to flag it as needing to be changed.

cmdutil: migrate `opts` on commitstatus() to native kwargs

cat: avoid round tripping **opts -> byteskwargs -> strkwargs Some internal users still want byteskwargs, so they are serviced in place. Once this pattern of changing types is eliminated everywhere, the remaining internal uses can be cleaned up individually (hopefully).

gpg: fix an UnboundLocalError whenever using --force It looks like this has been broke for almost a decade, since 9c89ac99690e.

rhg: fix the bug where sparse config is interpreted as relglob instead of glob relglob apparently (in contrast with relpath) matches everywhere in the tree, whereas glob only matches at the root. The python version interprets these patterns as "glob" (see "normalize(include, b'glob', ...)" in match.py)

hgweb: encode WSGI environment using the ISO-8859-1 codec The WSGI specification (PEP 3333) specifies that on Python 3 all strings passed by the server must be of type str with code points encodable using the ISO 8859-1 codec. For some reason, I introduced a bug in 2632c1ed8f34 by applying the reverse change. Maybe I got confused because PEP 3333 says that arbitrary operating system environment variables may be contained in the WSGI environment and therefore we need to handle the WSGI environment variables like we would handle operating system environment variables. The bug mentioned in the previous paragraph and fixed by this changeset manifested e.g. in the path of the URL being encoded in the wrong way. Browsers encode non-ASCII bytes with the percent-encoding. WSGI servers will decode the percent-encoded bytes and pass them to the application as strings where each byte is mapped to the corresponding code point with the same ordinal (i.e. it is decoded using the ISO-8859-1 codec). Mercurial uses the bytes type for these strings (which makes much more sense), so we need to encode it again using the ISO-8859-1 codec. If we use another codec, it can result in nonsense.

branching: merge stable into default

Added signature for changeset 787af4e0e8b7

Added tag 6.5.1 for changeset 787af4e0e8b7

relnotes: add 6.5.1

perf: fix perf::tags That return is invalid and unnecessary.

revlog: fix the naming scheme use by split temporary file The `-s` is now added on the first piece only and the `.i` is added to the index. This match the initially intended naming scheme.

revlog: show that temporary split file use broken name scheme The `-s` is added on the wrong part and the `.i` is missing.

fncache: fix a bug that corrupts the fncache after transaction rollback

fncache: add a test demonstrating fncache corruption

transaction: actually delete file created during the transaction on rollback Transaction currently has two modes: - one where file created during the transaction are deleted on rollback, - one where file created during the transaction are truncated to 0 on rollback. Before this change, `hg rollback` and `hg recover` are using the "delete" mode and transaction abort is using the "truncate" option. This difference is never really explained. A long time ago, there was two code paths, with this divergence existing for unclear reasons. When the two code paths got merged into a single one, a boolean argument have been added to preserve this divergence, mostly probably as a cargo cult. The divergence is weird and induce bad surprises, and the truncate behavior is a bit odds, introducing other bad surprises (e.g. 08ecbdba186f) So solve this, we stop using the "truncate" behavior and unify on the "delete" behavior. Despite being currently more "common", the truncate behavior seems less natural, resulting in the transaction leaving empty file around. This is landed on default, early in the cycle, to help us catch problems that could emerge.

rhg: tweak rhg fallback code structure The goal here is to share the code that accesses the config setting, but also move it to the top-level, so that we can more easily change this code later. (in fact in Jane Street we want to make it choose chg in some cases)