wireprotoserver: check permissions in main dispatch function The permissions checking code merged from stable is out of place in the refactored hgweb_mod module. This commit moves the main call to wireprotoserver. We still have some lingering code in hgweb_mod. This will get addressed later. Differential Revision: https://phab.mercurial-scm.org/D2717

wireprotoserver: check if command available before calling it The previous behavior was just plain wrong. I have no clue how it landed. My guess is a merge conflict resolution gone wrong on my end a few weeks ago. Differential Revision: https://phab.mercurial-scm.org/D2716

py3: drop encoding.strio() Its buffered nature makes TextIOWrapper unsuitable for temporarily wrapping bytes I/O.

ui: adjust Windows workaround to new _readline() code It's only needed when rawinput() is called. Also made it Py3 compatible.

ui: do not use rawinput() when we have to replace sys.stdin/stdout See the inline comment for why. The current Python3 hack doesn't work if more than one user inputs are expected because TextIOWrapper fills its internal buffer at the first read() request. Maybe we could write an unbuffered TextIOWrapper, but I don't want to make things more complicated. Instead, this patch reinvents raw_input(' ') of no readline support.

ui: do not try readline support if fin/fout aren't standard streams It's unlikely for a non-stdio stream to be a tty. Minimizing readline support makes it much simpler to work around the unicode input() function of Python 3. This also works on chg which duplicates client's tty to stdio fds.

util: add public isstdin/isstdout() functions

ui: add debug commands to test interactive prompt Interactive operations aren't easily covered by tests. So let's add commands to test them manually.

ui: inline util.bytesinput() into ui._readline() Prepares for rework of Python 3 support, which is currently broken due to read-ahead buffer of TextIOWrapper.

hgk: stop using util.bytesinput() to read a single line from stdin Appears that the stdio here is an IPC channel between hg and hgk (tk) processes, which shouldn't need a fancy readline support.

bookmarks: test for exchanging long bookmark names (issue5165) As far as I can tell a test for a long bookmark name never actually got added when this was fixed. Let's document that a 300 byte bookmark is something we're supporting (this patch started out life over a year ago as a way for me to validate the problem, and I recently found it.) I think the nonzero exits from the push operations are only because no new changesets get exchanged. Please correct me if I'm wrong on that. :) Differential Revision: https://phab.mercurial-scm.org/D2727

phabricator: follow-up phab auth improvements with backwards compat mode We'll rip this out before we ship 4.6, but this gives people a window to migrate. Differential Revision: https://phab.mercurial-scm.org/D2703

phabricator: specify API tokens per host, rather than per repo Differential Revision: https://phab.mercurial-scm.org/D1919

py3: drop b'' from generate-working-copy-states.py output I could make everything bytes, but decoding filename seemed easier and we don't have to worry about non-ascii filenames here.

py3: make test-commit-multiple.t byte-safe

py3: fix type of default username

py3: read/write plain lock file in binary mode A lock file shouldn't contain '\n', so this isn't a BC.

util: stop calling os.stat_float_times() It had Python-wide side effects, and it disappears in 3.7.0. As of this change, we're mostly working on 3.7.0b2. There are a few worrying failures, mostly around regular expressions, but we'll have to tackle those separately. Differential Revision: https://phab.mercurial-scm.org/D2697

cleanup: use stat_result[stat.ST_MTIME] instead of stat_result.st_mtime The latter is floating point by default, and we've been doing os.stat_float_times(False). Unfortunately, os.stat_float_times was removed between Python 3.7.0a1 and 3.7.0b2, so we have to stop using it. Differential Revision: https://phab.mercurial-scm.org/D2696

osutil: implement minimal __getitem__ compatibility on our custom listdir type We previously declined to do this, but the removal of the deprecated os.stat_float_times() method in Python 3.7 forces our hand. Differential Revision: https://phab.mercurial-scm.org/D2695

hgweb: adapt to socket._fileobject changes in Python 3 Differential Revision: https://phab.mercurial-scm.org/D2688

debugcommands: fix some %r output with bytestr() wrappers Almost fixes test-merge-tools.t. I think the remaining failure there is due to some overspecified tempfile names. Differential Revision: https://phab.mercurial-scm.org/D2675

tests: add test for issue 5494 but with --collapse This was not fixed, so the test case currently demonstrates the breakage. Differential Revision: https://phab.mercurial-scm.org/D2714

tests: .hg/merge is a directory, so use `test -d` This part of test-rebase-interrupts.t would have passed before the fix in a580b2d65ded (rebase: make sure merge state is cleaned up for no-op rebases (issue5494), 2017-05-18). Differential Revision: https://phab.mercurial-scm.org/D2713

rebase: only store collapse message once The message is determined by the user passing --message or --log when the rebase is started. There's no need to write it to a file for each rebased commit; writing it once at the start of the rebase is enough. Differential Revision: https://phab.mercurial-scm.org/D2712

rebase: collapse two nested if-conditions Also change the order since it feel to me like it's more about --collapse than it is about --keep. Differential Revision: https://phab.mercurial-scm.org/D2711

rebase: reduce scope of "dsguard" variables a bit Differential Revision: https://phab.mercurial-scm.org/D2710

rebase: remove unused argument "state" from rebasenode() Differential Revision: https://phab.mercurial-scm.org/D2709

rebase: delete obsolete internal "keepopen" option The option was apparently introduced for use by the "pbranch" extension, see f2558a8228be (rebase: add option to not commit after a collapsing, 2010-02-07). However, it doesn't seem like it was ever used by that extension (according to `hg grep` in a clone of https://bitbucket.org/parren/hg-pbranch/), so let's delete it. Differential Revision: https://phab.mercurial-scm.org/D2708

releasenotes: allow notes for multiple directives in a single changeset This problem was caught in da91e7309daf8ffc51bf3e6f4b2d8a16ef5af95a. This patch just makes sure there is no warning when we encounter such a case. Differential Revision: https://phab.mercurial-scm.org/D2254

releasenotes: mention changeset with warning and abort Output the changeset hash with the warning/abort message just to know where things messed up. Differential Revision: https://phab.mercurial-scm.org/D2253

releasenotes: replace abort with warning while parsing (issue5775) During the 4.5 development cycle, the extension broke on two different changesets. This fixes the issue by ensuring that it just throws a warning when it encounters unexpected behaviour, instead of aborting. Differential Revision: https://phab.mercurial-scm.org/D2255

archival: fileit should not use atomictemp, causes performance regression Differential Revision: https://phab.mercurial-scm.org/D2704

perf: teach perfbdiff to call blocks() and to use xdiff Differential Revision: https://phab.mercurial-scm.org/D2624

fuzz: fix xdiff build Recent xdiff code cleanups removed some files and changed some structures. Update fuzz code. Differential Revision: https://phab.mercurial-scm.org/D2707

xdiff: remove xmerge related logic hg has its own merge algorithm with flexible config options. Differential Revision: https://phab.mercurial-scm.org/D2706

xdiff: remove xemit related logic xemit handles "diff formatting and output" with options like context lines, whether show function names, etc. That is handled more cleanly at a higher level in hg. Removing context line parameters would also make the trimming logic (D2686) cleaner and more confident. See [1]. [1]: https://github.com/git/git/commit/d2f82950a9226ae1102a7a97f03440a4bf8c6c09 Differential Revision: https://phab.mercurial-scm.org/D2705

xdiff: remove unused structure, functions, and constants `bdiffparam_t` is unused. `xdl_fall_back_diff` is no longer used after D2573. `XDL_MMB_READONLY`, `XDL_MMF_ATOMIC` are unused. `XDL_BDOP*` are unused since there is no xdiff binary diff algorithm. `anchors` feature is not used. It's also relatively new in git. Differential Revision: https://phab.mercurial-scm.org/D2684

xdiff: remove whitespace related feature In Mercurial, whitespace related handling are done at a higher level than the low-level diff algorithm so "ignore spaces". So it's not used by mdiff. Some of the upcoming optimizations would be more difficult with whitespace related features kept. So let's remove them. Differential Revision: https://phab.mercurial-scm.org/D2683

merge with stable There were a handful of merge conflicts in the wire protocol code due to significant refactoring in default. When resolving the conflicts, I tried to produce the minimal number of changes to make the incoming security patches work with the new code. I will send some follow-up commits to get the security patches better integrated into default.

Added signature for changeset 8bba684efde7

Added tag 4.5.2 for changeset 8bba684efde7

merge with security patches

hgweb: always perform permissions checks on protocol commands (BC) (SEC) Previously, the HTTP request handling code would only perform permissions checking on a wire protocol command if that wire protocol command defined its permissions / operation type. This meant that commands (possibly provided by extensions) not defining their operation type would bypass permissions check. This could lead to exfiltration of data from servers and mutating repositories that were supposed to be read-only. This security issue has been present since the permissions table was introduced by d3147b4e3e8a in 2008. This commit changes the behavior of the HTTP server to always perform permissions checking for protocol requests. If an explicit permission for a wire protocol command is not defined, the server assumes the command can be used for writing and governs access accordingly. .. bc:: Wire protocol commands not defining their operation type in ``wireproto.PERMISSIONS`` are now assumed to be used for "push" operations and access control to run those commands is now enforced accordingly.

wireproto: check permissions when executing "batch" command (BC) (SEC) For as long as the "batch" command has existed (introduced by bd88561afb4b and first released as part of Mercurial 1.9), that command (like most wire commands introduced after 2008) lacked an entry in the hgweb permissions table. And since we don't verify permissions if an entry is missing from the permissions table, this meant that executing a command via "batch" would bypass all permissions checks. The security implications are significant: a Mercurial HTTP server would allow writes via "batch" wire protocol commands as long as the HTTP request were processed by Mercurial and the process running the Mercurial HTTP server had write access to the repository. The Mercurial defaults of servers being read-only and the various web.* config options to define access control were bypassed. In addition, "batch" could be used to exfiltrate data from servers that were configured to not allow read access. Both forms of permissions bypass could be mitigated to some extent by using HTTP authentication. This would prevent HTTP requests from hitting Mercurial's server logic. However, any authenticated request would still be able to bypass permissions checks via "batch" commands. The easiest exploit was to send "pushkey" commands via "batch" and modify the state of bookmarks, phases, and obsolescence markers. However, I suspect a well-crafted HTTP request could trick the server into running the "unbundle" wire protocol command, effectively performing a full `hg push` to create new changesets on the remote. This commit plugs this gaping security hole by having the "batch" command perform permissions checking on each sub-command that is being batched. We do this by threading a permissions checking callable all the way to the protocol handler. The threading is a bit hacky from a code perspective. But it preserves API compatibility, which is the proper thing to do on the stable branch. One of the subtle things we do is assume that a command with an undefined permission is a "push" command. This is the safest thing to do from a security perspective: we don't want to take chances that a command could perform a write even though the server is configured to not allow writes. As the test changes demonstrate, it is no longer possible to bypass permissions via the "batch" wire protocol command. .. bc:: The "batch" wire protocol command now enforces permissions of each invoked sub-command. Wire protocol commands must define their operation type or the "batch" command will assume they can write data and will prevent their execution on HTTP servers unless the HTTP request method is POST, the server is configured to allow pushes, and the (possibly authenticated) HTTP user is authorized to perform a push.

wireproto: declare operation type for most commands (BC) (SEC) The permissions model of hgweb relies on a dictionary to declare the operation associated with each command - either "pull" or "push." This dictionary was established by d3147b4e3e8a in 2008. Unfortunately, we neglected to update this dictionary as new wire protocol commands were introduced. This commit defines the operations of most wire protocol commands in the permissions dictionary. The "batch" command is omitted because it is special and requires a more complex solution. Since permissions checking is skipped unless a command has an entry in this dictionary (this security issue will be addressed in a subsequent commit), the practical effect of this change is that various wire protocol commands now HTTP 401 if web.deny_read or web.allow-pull, etc are set to deny access. This is reflected by test changes. Note how various `hg pull` and `hg push` operations now fail before discovery. (They fail during the initial "capabilities" request.) This change fixes a security issue where built-in wire protocol commands would return repository data even if the web config were configured to deny access to that data. I'm on the fence as to whether we should HTTP 401 the capabilities request. On one hand, it can expose repository metadata and can tell callers things like what version of Mercurial the server is running. On the other hand, a client may need to know the capabilities in order to authenticate in a follow-up request. It appears that Mercurial clients handle the HTTP 401 on *any* protocol request, so we should be OK sending a 401 for "capabilities." But if this causes problems, it should be possible to allow "capabilities" to always work. .. bc:: Various read-only wire protocol commands now return HTTP 401 Unauthorized if the hgweb configuration denies read/pull access to the repository. Previously, various wire protocol commands would still work and return data if read access was disabled.

wireproto: move command permissions dict out of hgweb_mod The operation type associated with wire protocol commands is supposed to be defined in a dictionary so it can be used for permissions checking. Since this metadata is closely associated with wire protocol commands themselves, it makes sense to define it in the same module where wire protocol commands are defined. This commit moves hgweb_mod.perms to wireproto.PERMISSIONS and updates most references in the code to use the new home. The old symbol remains an alias for the new symbol. Tests pass with the code pointing at the old symbol. So this should be API compatible for extensions. As part of the code move, we split up the assignment to the dict so it is next to the @wireprotocommand. This reinforces that a @wireprotocommand should have an entry in this dict. In the future, we'll want to declare permissions as part of the @wireprotocommand decorator. But this isn't appropriate for the stable branch.

tests: comprehensively test HTTP server permissions checking We didn't have test coverage for numerous web.* config options. We add that test coverage. Included in the tests are tests for custom commands. We have commands that are supposedly read-only and perform writes and a variation of each that does and does not define its operation type in hgweb_mod.perms. The tests reveal a handful of security bugs related to permissions checking. Subsequent commits will address these security bugs.

tests: extract HTTP permissions tests to own test file We're about to implement a lot more coverage of the permissions mechanism. In preparation for that, establish a new test file to hold permissions checks. As part of this, we inline the important parts of the "req" helper function.

Added signature for changeset 369aadf7a326

Added tag 4.5.1 for changeset 369aadf7a326

revlog: resolve lfs rawtext to vanilla rawtext before applying delta This happens when a LFS delta base gets a non-LFS delta from another client. In that case, the LFS delta base needs to be converted to non-LFS version before applying the delta. Differential Revision: https://phab.mercurial-scm.org/D2069

sslutil: some more forcebytes() on some exception messages At this point, test-https.t no longer dumps tracebacks everywhere. Instead, we get some results that look like we're not adequately finding things in hg's configuration, which should be manageable (if somewhat annoying to find and fix.) Differential Revision: https://phab.mercurial-scm.org/D2690

sslutil: sslcontext needs the cipher name as a sysstr Differential Revision: https://phab.mercurial-scm.org/D2689

sslutil: lots of unicode/bytes cleanup In general, we handle hostnames as bytes, except where Python forces them to be unicodes. This fixes all the tracebacks I was seeing in test-https.t, but there's still some ECONNRESET weirdness that I can't hunt down... Differential Revision: https://phab.mercurial-scm.org/D2687

revlog: do not use delta for lfs revisions This is similar to what we have done for changegroups. It is needed to make sure the delta application code path can assume deltas are always against vanilla (ex. non-LFS) rawtext so the next fix becomes possible. Differential Revision: https://phab.mercurial-scm.org/D2068

changegroup: do not delta lfs revisions There is no way to distinguish whether a delta base is LFS or non-LFS. If the delta is against LFS rawtext, and the client trying to apply it has the base revision stored as fulltext, the delta (aka. bundle) will fail to apply. This patch forbids using delta for LFS revisions in changegroup so bad deltas won't be transmitted. Note: this does not solve the problem entirely. It solves LFS delta applying to non-LFS base. But the other direction: non-LFS delta applying to LFS base is not solved yet. Differential Revision: https://phab.mercurial-scm.org/D2067

lfs: add a test showing bundle application could be broken When a bundle containing LFS delta uses non-LFS delta-base, or vice-versa, the bundle will fail to apply. Differential Revision: https://phab.mercurial-scm.org/D2066

debugwireproto: handle unimplemented util.poll() for Windows This is the same logic used in sshpeer.doublepipe. It doesn't completely fix test-ssh-proto{,-unbundle}.t ("read(-1) -> X" is changed to "read(X) -> X", the order of some lines are changed, and abort messages seem to be missing), but it cuts down a ton on the failure spew.

py3: byte-stringify test-blackbox.t

py3: byte-stringify blackbox-readonly-dispatch.py # skip-blame because just adding some b''

py3: make blackbox-readonly-dispatch.py use ui instead of print()

py3: fix int formatting of "incoming changes" log

largefiles: use %d instead of %s to process ints Differential Revision: https://phab.mercurial-scm.org/D2677

transaction: fix an error string with bytestr() on a repr()d value Fixes test-rollback.t on Python 3. Differential Revision: https://phab.mercurial-scm.org/D2674

py3: work around comparison between int and None in tagmerge

py3: do not mutate dict while iterating in tagmerge

py3: fix type of ui.configitems(ignoresub=True) result

py3: don't use str() to stringify pushloc

py3: byte-stringify test-config.t and test-config-env.py

py3: use startswith() instead of slicing to detect leading whitespace

archival: use py3 friendly replacements for chr() and long() Differential Revision: https://phab.mercurial-scm.org/D2673

archival: ensure file mode for gzipfile is sysstr Differential Revision: https://phab.mercurial-scm.org/D2672

archival: fix a missing r'' on a kwargs check # skip-blame just an r prefix Differential Revision: https://phab.mercurial-scm.org/D2671

py3: more passing tests (ten this time) Differential Revision: https://phab.mercurial-scm.org/D2670

util: fix unsafe url abort with bytestr() on url Differential Revision: https://phab.mercurial-scm.org/D2669

xdiff: fix builds on Windows This works on my ancient Fedora system too, without warnings. There are, however, warnings about various 64 to 32 bit conversions on Windows that need to be examined.

py3: use bytes() instead of str() Differential Revision: https://phab.mercurial-scm.org/D2665

py3: use b"%d" instead of str() to convert integers to bytes Differential Revision: https://phab.mercurial-scm.org/D2664

py3: use util.forcebytestr to convert testedwith value to bytes Bad extensions can put anything in testedwith so we should use util.forcebytestr here. Differential Revision: https://phab.mercurial-scm.org/D2663

py3: use pycompat.bytestr instead of str Differential Revision: https://phab.mercurial-scm.org/D2648

setdiscovery: avoid a Yoda condition Differential Revision: https://phab.mercurial-scm.org/D2646

setdiscovery: remove unnecessary sample size limiting Both _takequicksample() and _takefullsample() already limit their result to the request size, so there's no need to let the caller do that again. Differential Revision: https://phab.mercurial-scm.org/D2645

setdiscovery: remove initialsamplesize from a condition It seems more direct to compare the actual sample size. That way we can change the sample taken earlier in the code without breaking the condition. Differential Revision: https://phab.mercurial-scm.org/D2644

setdiscovery: back out changeset 5cfdf6137af8 (issue5809) As explained in the bug report, this commit caused a performance regression. The problem occurs when the local repo has very many heads. Before 5cfdf6137af8, we used to get the remote's list of heads and if these heads mostly overlapped with the local repo's heads, we would mark these common heads as common, which would greatly reduce the size of the set of undecided nodes. Note that a similar problem existed before 5cfdf6137af8: If the local repo had very many heads and the server just had a few (or many heads from a disjoint set), we would do the same kind of slow discovery as we would with 5cfdf6137af8 in the case where local and remote repos share a large set of common nodes. For now, we just back out 5cfdf6137af8. We should improve the discovery in the "local has many heads, remote has few heads" case, but let's do that after backing this out. Differential Revision: https://phab.mercurial-scm.org/D2643

hgweb: fix up trailing slash detection on Python 3 Fixes a couple of hgweb tests. Differential Revision: https://phab.mercurial-scm.org/D2661

hgweb: convert req.form to bytes for all keys and values This is just going to be a lot cleaner for our internals. Differential Revision: https://phab.mercurial-scm.org/D2660

templater: show repr of string we're rejecting I feel like this should make it a little easier to hunt down problems. Differential Revision: https://phab.mercurial-scm.org/D2659

webutil: some %d instead of %s love on ints Differential Revision: https://phab.mercurial-scm.org/D2658

py3: whitelist three more cases Differential Revision: https://phab.mercurial-scm.org/D2657

archival: our filenames are bytes, not strs Differential Revision: https://phab.mercurial-scm.org/D2656

archival: tar file modes need to be sysstrs Differential Revision: https://phab.mercurial-scm.org/D2655

archival: fsdecode paths before passing to tar or zip objects Both of these traffic in unicodes for filenames on Python 3, and inspection of the tarfile module shows that it uses the filesystem encoding, so fsdecode is the right choice. Differential Revision: https://phab.mercurial-scm.org/D2654

py3: add b'' prefixes in tests/test-minirst.py # skip-blame because just b'' prefixes Differential Revision: https://phab.mercurial-scm.org/D2653

py3: make sure __repr__ returns a str # skip-blame because just r'' prefix Differential Revision: https://phab.mercurial-scm.org/D2652

py3: make sure regular expressions are bytes # skip-blame because just b'' prefix Differential Revision: https://phab.mercurial-scm.org/D2651

py3: use bytes instead of str to make sure we use bytes internally Differential Revision: https://phab.mercurial-scm.org/D2650

py3: use util.forcebytestr instead of str to convert error messages Differential Revision: https://phab.mercurial-scm.org/D2649

lock: block signal interrupt while making a lock file On Windows where symlink isn't supported, util.makelock() could leave an empty file if interrupted immediately after os.open(). This empty lock never dies as it has no process id recorded. ld = os.open(pathname, os.O_CREAT | os.O_WRONLY | os.O_EXCL) # an interrupt may occur here os.write(ld, info) os.close(ld) This was a long-standing bug of TortoiseHg which runs a command-server and kills it by CTRL_C_EVENT, reported by random Windows users. https://bitbucket.org/tortoisehg/thg/issues/4873/#comment-43591129 At first, I tried to fix makelock() to clean up a stale lock file, which turned out to be hard because any instructions may be interrupted by a signal. ld = None try: # CALL_FUNCTION # os.open(...) # an interrupt may occur here # STORE_FAST # ld = ... ld = os.open(pathname, os.O_CREAT | os.O_WRONLY | os.O_EXCL) os.write(ld, info) ... return True except: if ld: ... os.unlink(pathname) return False So I decided to block signals by temporarily replacing the signal handlers so makelcok() and held = 1 will never be interrupted. Many thanks to Fernando Najera for investigating the issue.

fuzz: add some more docs about building/running fuzzers Differential Revision: https://phab.mercurial-scm.org/D2635

util: also silence py3 warnings from codec module Fixes warnings like this: + mercurial/util.py:2446: DeprecationWarning: invalid escape sequence '\d' + return codecs.escape_decode(s)[0] Differential Revision: https://phab.mercurial-scm.org/D2642

test-annotate: set stdin and stdout to binary to get CR unmodified

test-annotate: rewrite sed with some python I hope this will fix the test failure seen on FreeBSD and Windows.

docs: small fixes for profiling.nested and the overall description - profiling.nested defaults to 0, not 5 - profiling is not always done with lsprof Differential Revision: https://phab.mercurial-scm.org/D2641

scmutil: fix oversight in b76248e51605c6 where I forgot to use msg Thanks to Yuya for spotting my mistake. Differential Revision: https://phab.mercurial-scm.org/D2636

hghave: remove unused "as ex" in exception block I overlooked this when removing a debug print in another change. Differential Revision: https://phab.mercurial-scm.org/D2637

tests: port test-log to Python 3 Required some porting to >>> inline Python instead of using heredocs into $PYTHON. Differential Revision: https://phab.mercurial-scm.org/D2621

py3: make gettext domain a system string

templater: fix position of terminator character in error message Since a template expression starts after the '{' character, the expression should be considered to end immediately before the terminator '{'.

merge with stable

test-subrepo: glob away an unstable hash This is the instability mentioned at the beginning of the series. I don't like hiding it, but I don't want to sit on a fix for a user reported problem while trying to figure this out. The instability seems related to the cset with a .hgsub with a remote URL. (There's very little existing remote URL subrepo testing.)

subrepo: activate clone pooling to enable sharing with remote URLs This is the easiest way to ensure that repositories with remote subrepo references can share the subrepos, consistent with how local subrepos can be shared.

subrepo: don't attempt to share remote sources (issue5793) Untangling _abssource() to resolve the new subrepo relative to the shared parent's share path, and then either sharing from there (if it exists), or cloning to that location and then sharing, is probably more than should be attempted on stable. Absolute subrepo references are discouraged, so for now, this resumes the behavior prior to 68e0bcb90357 of cloning the absolute subrepo locally.

test-subrepo: demonstrate problems with subrepo sharing and absolute paths This affects remote paths in .hgsub, as well as clone pooling from a remote source. For reasons unknown, there are stability issues with the relative-path.t tests. If run as a single test, it is stable. If run with --loop, or with -jX for X>1, the hash of the parent repo changes. I'm seeing this on both Windows and Fedora 26. I added an `hg log --debug`, and the manifest hash changes, but I have no idea why.

revsetlang: add a hint for more useful parse errors This logic is largely based on the similar logic added to template error messages in D2608 and D2609, but with a few tweaks based on how revsets actually work. Differential Revision: https://phab.mercurial-scm.org/D2619

setup: ignore extension load failures when finding working hg Previously, `make local` would fail if any extension was not properly loading. Differential Revision: https://phab.mercurial-scm.org/D2589

profile: colorize output on Windows Previously, the ANSI codes were printed to the screen, throwing off the alignment. We could probably do this unconditionally, but it's kind of a hack, so I figured I'd limit the scope.

dispatch: don't clamp the range of the exit code twice We already limit the range to (0, 255) in the call to sys.exit(). The duplicated operation can't possibly be hurting us, but let's clean it up to avoid confusion.

bdiff: avoid pointer arithmetic on void* MSVC 2008 complains: mercurial/cext/bdiff.c(106) : error C2036: 'void *' : unknown size mercurial/cext/bdiff.c(107) : error C2036: 'void *' : unknown size Maybe it's a gcc extension? https://stackoverflow.com/questions/37460579/error-c2036-void-unknown-size

fuzz: add a quick README to try and document how to test new fuzzers Differential Revision: https://phab.mercurial-scm.org/D2633