py3: make sure we return str from __repr__ Differential Revision: https://phab.mercurial-scm.org/D2109

py3: check for bytes instead of str in hg.share() Differential Revision: https://phab.mercurial-scm.org/D2151

py3: port metaedit extension to Python 3 Without these changes, this extension is responsible for a test of test failures. Differential Revision: https://phab.mercurial-scm.org/D2150

py3: pass system string to email.message.Message.set_type() Python 3 insists the type is a str. Differential Revision: https://phab.mercurial-scm.org/D2149

py3: open temporary file in binary mode Otherwise things fail later when we write bytes to the handle. Differential Revision: https://phab.mercurial-scm.org/D2148

py3: use email parser that operates on bytes email.parser.Parser() operates on str in both Python 2 and 3. Python 3.2 introduced the email.parser.BytesParser(), which works like Parser except it accepts bytes. We implement the pycompat helper as a function so we lazily import the "email" module. Differential Revision: https://phab.mercurial-scm.org/D2147

py3: port ext-phase-report.py extension The custom module importer doesn't run on Python files in the tests directory. So we need the source to be compatible with both Python 2 and 3. Differential Revision: https://phab.mercurial-scm.org/D2145

py3: more robustly cast UUID to bytes The UUID type only has __str__ implemented. So we need to cast to bytes on Python 3. We need an actual bytes instance here (bytestr won't do) because the re.escape() later iterates over characters and characters need to behave like ints, not bytes instances of length 1. Differential Revision: https://phab.mercurial-scm.org/D2144

py3: use bytes() to cast context instances __str__ and __bytes__ are both implemented on context types. However, __str__ behaves differently on Python 2 and 3. Differential Revision: https://phab.mercurial-scm.org/D2143

py3: convert __doc__ to bytes Differential Revision: https://phab.mercurial-scm.org/D2142

py3: cast repr() result to bytes Differential Revision: https://phab.mercurial-scm.org/D2141

py3: fix file i/o in test-status.t Use raw string literal for open() mode and pass the proper data type to write(). Differential Revision: https://phab.mercurial-scm.org/D2140

py3: pass system string to socket.getservbyname Differential Revision: https://phab.mercurial-scm.org/D2139

run-tests: report tests that exception occurred in We now record the test that an exception occurred in. We put this information to use by aggregating the count of failures in each test. For each exception, the exception report now prints the total number of tests having that exception and the test with the least number of exceptions exhibiting that failure. The exception list is now sorted by (total count, tests impacted, count of failures in least failing test). This allows us to: * Assess how widespread a failure is. Some exceptions occur a lot in a few tests. Others occur over many tests. * Easily run a test exhibiting an exception without having to find a failure in test output. * Find and fix low hanging fruit (e.g. exceptions that are the only failure in a test). Here's an example of the new output: 199 (4 tests) /home/gps/src/hg/hgext/blackbox.py:191: %b requires a bytes-like object, or an object that implements __bytes__, not 'str' (test-devel-warnings.t - 1 total) 142 (19 tests) /home/gps/src/hg/hgext/mq.py:655: list indices must be integers or slices, not bytes (test-hardlinks.t - 1 total) 140 (20 tests) /home/gps/src/hg/mercurial/patch.py:296: string argument expected, got 'bytes' (test-audit-subrepo.t - 1 total) 101 (15 tests) /home/gps/src/hg/hgext/convert/convcmd.py:60: encode() argument 1 must be str, not bytes (test-convert-clonebranches.t - 1 total) 90 (2 tests) /home/gps/src/hg/hgext/mq.py:456: can't concat str to bytes (test-mq-qqueue.t - 1 total) 87 (2 tests) /home/gps/src/hg/mercurial/branchmap.py:380: %b requires a bytes-like object, or an object that implements __bytes__, not 'FileNotFoundError' (test-branches.t - 2 total) 85 (22 tests) /home/gps/src/hg/mercurial/sshpeer.py:223: cannot convert 'UUID' object to bytes (test-bundle2-pushback.t - 1 total) 1 (1 tests) /home/gps/src/hg/mercurial/formatter.py:254: %b requires a bytes-like object, or an object that implements __bytes__, not 'str' (test-debugextensions.t - 2 total) 1 (1 tests) /home/gps/src/hg/hgext/convert/convcmd.py:420: startswith first arg must be str or a tuple of str, not bytes (test-convert-authormap.t - 2 total) 1 (1 tests) /home/gps/src/hg/mercurial/revlog.py:797: '>=' not supported between instances of 'NoneType' and 'int' (test-unionrepo.t - 1 total) 1 (1 tests) /home/gps/src/hg/hgext/show.py:129: %b requires a bytes-like object, or an object that implements __bytes__, not 'str' (test-show.t - 1 total) Differential Revision: https://phab.mercurial-scm.org/D2138

py3: use raw string for file open mode Otherwise the module importer rewrites it to bytes on Python 3, which angers Python 3. Differential Revision: https://phab.mercurial-scm.org/D2137

py3: convert content-type to bytes This fixes the exception: TypeError: %b requires a bytes-like object, or an object that implements __bytes__, not 'str' On the line: ui.debug('Content-Type: %s\n' % content_type) And it prevents additional exceptions that use the content_type variable later. Differential Revision: https://phab.mercurial-scm.org/D2136

gpg: print unknown key IDs in their entirety Shortening the key is nice in theory but it results in ambiguity which can be exploited. Therefore, when encountering an unknown key ID we should print the whole ID returned by gpg. This may or may not be the whole key, however it will match the user preference set in gpg configuration. Furthermore, the key ID shortening had a couple of issues: (1) it truncated the key ID (dropping the last digit and outputting only 15 hex digits) making it very hard to find the correct key on a key server (2) since only 15 digits were fed into shortkey(), it always emitted the ui.debug() warning

tests: fix check for --graph option in test-log-linerange.t Previous test about incompatibility of --line-range and --graph options was problematic because, even without --graph option, the command would have failed because specified path (dir/baz) did not exist as it got dropped in previous test statements. New test statements should work without --graph. This prepares for making --line-range compatible with --graph option.

py3: use "%d" instead of "%s" for integers Differential Revision: https://phab.mercurial-scm.org/D2135

py3: replace file() with open() in test-contrib.t file() is not present in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2134

py3: replace file() with open() in test-eol.t file() is not present in Python 3. This patch also adds a b'' prefix to make sure we write bytes in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2129

py3: replace file() with open() in test-extdiff.t file() is not present in Python 3. This patch also adds a b'' prefix to make sure we write bytes in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2128

py3: replace file() with open() in test-fileset.t file() is not present in Python 3. This patch also adds a b'' prefix to make sure we write bytes in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2127

py3: replace file() with open() in test-import-bypass.t file() is not present in Python 3. This patch also adds a b'' prefix to make sure we write bytes in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2126

py3: replace file() with open() in test-import-eol.t file() is not present in Python 3. This patch also adds b'' prefixes to make sure we write bytes in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2125

py3: replace file() with open() in test-import-git.t file() is not present in Python 3. This patch also adds a b'' prefix to make sure we write bytes in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2124

py3: replace file() with open() in test-import.t file() is not present in Python 3. This patch also adds a b'' prefix to make sure we write bytes in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2123

py3: replace file() with open() in test-mactext.t file() is not present in Python 3. This patch also adds a b'' prefix to make sure we write bytes in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2122

py3: replace file() with open() in test-mq-missingfiles.t file() is not present in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2121

py3: replace file() with open() in test-mq-qimport.t file() is not present in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2120

py3: replace file() with open() in test-patch-offset.t file() is not present in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2119

py3: replace file() with open() in test-relink.t file() is not present in Python 3. This patch also adds a b'' prefix to make sure we write bytes in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2118

py3: replace file() with open() in test-subrepo-missing.t file() is not present in Python 3. This patch also adds a b'' prefix to make sure we write bytes in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2117

py3: replace file() with open() in test-tag.t file() is not present in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2116

py3: replace file() with open() in test-transplant.t file() is not present in Python 3. This patch also adds a b'' prefix to make sure we write bytes in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2115

py3: replace file() with open() in test-walk.t file() is not present in Python 3. This patch also adds a b'' prefix to make sure we write bytes in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2114

py3: replace file() with open() in test-win32text.t This patch also adds b'' prefix to make sure we write bytes on Python 3. Differential Revision: https://phab.mercurial-scm.org/D2113

py3: replace file() with open() in test-convert-p4-filetypes.t file() is not present in Python 3. While I was here, I added b'' prefix to make sure we write things correctly in Python 3. Differential Revision: https://phab.mercurial-scm.org/D2112

py3: replace file() with open() in test-convert-mtn.t While I was here, I added b'' prefixes to make sure we write things correctly in bytes mode on Python 3. Differential Revision: https://phab.mercurial-scm.org/D2111

py3: replace file() with open() in test-clone.t While I was here, I made sure we write things correctly in bytes mode. Differential Revision: https://phab.mercurial-scm.org/D2110

merge: cut import cycle at merge -> extensions This seems less bad because checking fsmonitor looks like a temporary hack, and cmdutil -> mergemod should be a valid dependency.

subrepo: split non-core functions to new module Resolves import cycle caused by subrepo -> cmdutil. Still we have another cycle, cmdutil -> context -> subrepo, but where I think importing context is wrong. Perhaps we'll need repo.makememctx().

diff: remove fp.write() wrapper which drops label argument It's no longer needed since we've split labeled write() from file-like write().

log: pack filematcher and hunksfilter into changesetdiffer object This is just a way of getting rid of clumsy makefilematcher/makehunksfilter arguments. There might be a better abstraction, but I don't think this is bad. This makes filematcher and hunksfilter available by default, but that should be fine.

log: drop dead code to concatenate --line-range patterns and pats It's disabled since 2e45bbd3db7b, and the current implementation is unlikely to be reused.

logcmdutil: create hunksfilter and filematcher even if no diff option given It's okay since 5fe6f946f111, "log: allow matchfn to be non-null even if both --patch/--stat are off."

logcmdutil: unindent diff generator of changesetprinter Prepares for the next few patches which will make matchfn and hunksfilterfn always available.

logcmdutil: hold makefilematcher/makehunksfilter() by changesetpriner (API) This merges self.matchfn and self.show(matchfn) into self._makefilematcher, and does the same for hunksfilter. Because changesetprinter seems to have too many optional arguments, makefilematcher() and makehunksfilter() will be packed into one object by later patch.

log: pass ctx to makefilematcher() and makehunksfilter() functions This isn't important, but seems more consistent as changesetprinter.show() takes a ctx, not a revision number.

lfs: teach the 'lfs()' fileset to handle removed files The callstatus setting is required to notice the removal of 'lfs.test' in rev 6 in the tests, even though this isn't directly calling mctx.status(). However, it's not needed to get the results in the tests for `hg status`, so I'm probably missing something.

lfs: teach '{lfs_files}' to handle removed files

lfs: allow a pointer to be extracted from a context that removes the file This is needed to let 'set:lfs()' and '{lfs_files}' work normally on removed files. Yuya suggested returning a null pointer for removed files, instead of the pointer from the parent. The first attempt at this was to return None for a non LFS file, and a (pointer, ctx) tuple to hold the pointer and context (or parent pointer and context for a removed file). But this complicated the callers, even the ones that didn't care about removed files. Instead, let's use {} to represent a removed pointer. This has the added convenience of being a useful representation in the template language, and only affects the callers that care about removed files (and only slightly). Since pointers are explicitly serialized with a call to a member function, there is no danger of writing these to disk.

rebase: make "successors" a set in _computeobsoletenotrebased() There's no apparent reason for this variable to be a list and this avoids converting it to a set when needed.

merge with stable

rebase: do not consider extincts for divergence detection (issue5782) Extinct obsolete changesets cannot cause divergence upon rebase. We compute these obsoletes without a non-obsolete successor (extincts) in _computeobsoletenotrebased() and then filter them out from the set of obsolete revisions to rebase before getting into _checkobsrebase() to check for divergence candidates.

rebase: eliminate node from successors early in _computeobsoletenotrebased()

rebase: add a test case for issue5782 Issue 5782 reports that rebase incorrectly aborts when trying to rebase an extinct revision (an obsolete revision with only obsolete successor). We add a test to demonstrate this: the first "hg rebase" command aborts with the divergence warning while, when allowing divergence, the rebase completes and does not actually produce any divergence.

test: glob the temporary directory out of temporary file path The temporary directory used by python might be outside of '/tmp/' (eg: /dev/shm/) so we glob that part out.

tests: raise a better error when patterns are wrongly formatted It is fairly easy to make mistakes when merging conflict in the pattern file. A common mistake is to forget adding an extra trailing comma changing the length of the tuple. We now detect such error and raise a better error message that helps to find it.

lfs: add a fileset for detecting lfs files This currently has the same limitation as {lfs_files}, namely it doesn't report removed files. We may want a dedicated 'lfs()' revset for efficiency, but combining this with the 'contains()' revset should be equivalent for now. Combining with 'set:added()' or 'set:modified()' inside 'files()' should be equivalent to a hypothetical lfs_adds() and lfs_modifies(). I wonder if there's a way to tweak the filesets to evaluate lazily, to close the efficiency gap. It would also be interesting to come up with a template filter for '{files}' that looked at the pattern to 'files()', and filtered appropriately. While passing a fileset as the pattern to `hg log` does filter '{files}', the set is evaluated against the working directory, so there's no way to list all non-lfs files above a certain size in all revisions, for example.