tests: demonstrate inconsistent messaging around interrupted rebases

tests: adapt test-check-pylint to run on Windows The line endings are explicitly converted because this was ending up with '\r (no-eol) (esc)' lines, in addition to the usual '\r (esc)' lines. I've seen the fakerc output on other recently installed systems though (10.13 and/or Fedora 26). Unfortunately, the output here uses '\\' on Windows, so glob away the whole path.

hghave: avoid a deadlock reading the child process's output The output of `pylint` is voluminous enough that it fills the buffer on Windows, and waited for the parent to read it out. But the parent was waiting on the child to exit. I'm not sure what the intent of setting `ret = -1` in the exception handler just above this was...

lfs: bypass wrapped functions when reposetup() hasn't been called (issue5902) There are only a handful of methods that access repo attributes that are applied in reposetup(). The `diff` test covers all of the commands that call scmutil.prefetchfiles(). Along the way, I saw that adding files and upgrading the repo format were also problems (also tested here). I don't think running `hg serve` through the commandserver is sane, but I conditionalized both the capabilities and the wsgirequest handler because it's trivially correct. It doesn't look like there has ever been a caller of candownload(), so there's no test for that path. The upload case isn't testable, because uploadblobs() bails if there are no pointers. The requirement should be added any time pointers are introduced, and that would force the extension to be loaded specifically for the repo. This covers `debuglfsupload`, the pre-push hook (which isn't set until the repo is promoted to LFS), and uploadblobsfromrevs(), which can be called by other extensions. I think readfromstore() and writetostore() are only reachable as a flag processor for revlog.REVIDX_EXTSTORED, and a requirement is added as soon as that is seen, so I don't think those are a problem.

revlog: disallow setting uncompressed length to None Backed out changeset a4942675de6b

unionrepo: fill in uncompressed length of revlog entry It can be either -1 or a valid length, but shouldn't be None. IIUC, we can simply trust the length retrieved from the revlog to be overlaid. I don't bother thinking whether the compressed length can be copied as well. We'll need to fix it later.

fuzz: add clean target Differential Revision: https://phab.mercurial-scm.org/D3677

fuzzutil: make it possible to use absl when C++17 isn't supported Differential Revision: https://phab.mercurial-scm.org/D3676

fuzz: extract some common utilities and use modern C++ idioms Alex Gaynor suggested we should probably copy the left and right sides of diffs to new blocks so we can detect over-reads in the diffing code, and I agree. Once I got into that, I realized we should do things with C++17 idioms rather than keep using malloc() and free(). This change is the result. I tried to split it more than this and failed. Everything still compiles and works in the oss-fuzz container, so I think we can count on C++17 being available! Differential Revision: https://phab.mercurial-scm.org/D3675

tests: update fuzzer tests to include both fuzzers Differential Revision: https://phab.mercurial-scm.org/D3674

transaction-summary: show phase changes statistics in pull/unbundle Upon pull or unbundle, we display a message with the number of changesets which phase became public. Noticeably, this new message would appear even if no new changeset were added (below the "no changes found" message), thus indicating that something actually happened to the local repository.

outgoing: pay attention to `default:pushurl` for bookmarks and subrepos The problem here was that `default:pushurl` and `default` get translated to a single entry in `ui.paths` named 'default', with an attribute for 'pushloc', 'loc', and 'rawloc'. ui.expandpath() then always takes the `rawloc` attribute. Maybe the ui.expandpath() API is busted and should be removed? Or maybe getpath() should return a copy that adds an attribute reflecting the URL of the path chosen? I thought that I could remove the code in hg._outgoing() and pass the location resolved in commands.py as `dest`, but unfortunately that code is needed there to resolve #branch type URLs. Maybe that should be pulled up to commands.py, because I can't see any reasonable behavior for a subrepo path that's constructed out of that type of URL. The push command already resolves this early, so that works properly. But it looks like bundle, histedit, largefiles, patchbomb, and summary use a similar pattern, so they are likely similarly affected.

revlog: make chainbase cache its result for the correct revision Previously, as 'rev' was our iterator, we were always caching the chain base for the second revision of the chain, or for the base itself.

revlog: make getcandidaterevs more consistent about updating tested revs set Like in previous cases, update the set of tested revisions after yielding

templatekw: make getrenamed() return only filename, not nodeid No callers cared about the nodeid and I want to make getrenamed() not look up the node (although it's currently free, I hope to store copy info in changesets and not include the nodeid). Differential Revision: https://phab.mercurial-scm.org/D3666

py3: wrap tempfile.NamedTemporaryFile() to return bytes fp.name Unlike its name, tempfile.NamedTemporaryFile is not a class, so I renamed the pycompat version to look like a plain function. Since temp.name uses in the infinitepush extension aren't bytes-safe, this patch leaves them unmodified. Another weird thing is tempfile.mktemp(), which does not accept bytes suffix nor prefix. Sigh.

py3: wrap tempfile.mkdtemp() to use bytes path This also flips the default to use a bytes path on Python 3.

py3: wrap tempfile.mkstemp() to use bytes path This patch just flips the default to use a bytes path on Python 3. ca1cf9b3cce7 is backed out as the bundlepath should be bytes now.

extensions: remove strip_init=True from _disabledpaths() It's no longer used.

extensions: peek command table of disabled extensions without importing With chg where demandimport disabled, and if disk cache not warm, it took more than 5 seconds to get "unknown command" error when you typo a command name. This is horrible UX. The new implementation is less accurate than the original one as Python can do anything at import time and cmdtable may be imported from another module, but I think it's good enough. Note that the new implementation has to parse .py files, which is slightly slower than executing .pyc if demandimport is enabled.

lfs: clarify pointer validation error messages It wasn't obvious that LFS was involved from the error messages when `hg verify` fails.

terse: pconvert() entries added to the temporary terse dict for Windows Recent additional testing revealed this problem on Windows: --- tests/test-status.t.err +++ tests/test-status.t.err @@ -109,7 +109,7 @@ tweaking defaults works $ hg status --cwd a --config ui.tweakdefaults=yes - ? . + ? ../a/ ? ../b/ ? ../in_root $ HGPLAIN=1 hg status --cwd a --config ui.tweakdefaults=yes @@ -120,7 +120,7 @@ ? b/in_b (glob) ? in_root $ HGPLAINEXCEPT=tweakdefaults hg status --cwd a --config ui.tweakdefaults=yes - ? . + ? ..\a\ ? ../b/ ? ../in_root (glob) AFAICT, the status list (input and output here) is always in '/' format. The '\' printed output on Windows is because each file is run through repo.pathto() -> dirstate.pathto() -> util.pathto(). (And that function states that the argument uses '/' separators.) I fixed a similar issue in 362096cfdb1f, and given the apparent need for these strings to be in '/' format, I wonder if cmdutil.dirnode() should be rewritten to avoid os.path.join(). But it looks like all entries added to the temporary terse dict should use '/' now, and cmdutil.tersedir() looks like the only user.

filemerge: don't pass function name as loadpath's module_name param

phases: define an official tuple of phases we do not share Same motivation as for `mutablephases`, having a single definition helps with updating phases logic.

phases: define an official tuple of mutable phases Such tuple was already manually defined in a couple of place. Having an official definition makes it easy to introduce of new phases.

repoview: use 'phasecache.getrevset' when computing 'unserved' filter This reuses a precomputed set, being much faster than manual iteration. Computing the "unserved" filter speed up by 55%. before: wall 0.014671 comb 0.020000 user 0.020000 sys 0.000000 (best of 180) after: wall 0.006623 comb 0.010000 user 0.010000 sys 0.000000 (best of 401) The "unserved" filter is used to restrict the revisions client can pull from a server. (eg: secret changesets)

graft: drop --user and --date values info from hint in case of conflicts After previous patch, we have started preserving user and date values in graftstate and reusing them during `hg graft --continue`. Now passing --user and --date again with --continue makes no sense. Let's drop them from the hint. Differential Revision: https://phab.mercurial-scm.org/D3660

graft: reuse --user and --date values in `hg graft --continue` (BC) Reading the user and date information from graftstate during `hg graft --continue` will help us in preserving the user and date arguments passed when `hg graft` was called. This patch reads that information and reuses that while running `hg graft --continue`. So after this patch, --user and --date values are preserved even if conflicts occur and user don't need to pass them again. The test changes demonstrate the fix. This is a backward incompatible change but I think of this more as a bug fix. Also thinking about removing the line from `hg help graft` which says --continue does not reapply other flags but need to check what are the other flags which needs to be preserved. Differential Revision: https://phab.mercurial-scm.org/D3659

tests: add test showing --continue not preserving --date and --user flags This patch adds test showing that we don't preserve the user passed --date and --user values in `hg graft`. I was fixing that and realized this is untested. Adding tests before so that behavior change or the fix is easy to realize. Differential Revision: https://phab.mercurial-scm.org/D3658

graft: store user passed date and user information in graftstate Right now, `hg help graft` says: The -c/--continue option does not reapply earlier options, except for --force. which should be treated as a bug. A good user experience is that the commands remember the arguments passed initially and preserve them during `hg graft --continue`. This patch starts storing the user and date information in graftstate if user passed it. Upcoming patches will make sure we preserve that information during --continue and them don't allow user to pass any new arguments with --continue. I don't think there is any another `--continue` flag which allows new options to be passed with it. Differential Revision: https://phab.mercurial-scm.org/D3657

graft: use cmdstate.delete() to delete the state file Previous patches start using state.cmdstate() class for statefiles. The class has a function delete() to delete the state file. This patch replaces the existing repo.vfs.unlink() with that delete() function. Differential Revision: https://phab.mercurial-scm.org/D3656

graph: add outputgraph() function, called by ascii() to print the graph to the ui. This allows a cleaner entrypoint for extensions to tweak the graph output without needing to rewrite all of ascii(), or needing to manually guess where the graph nodes/edges end and the rev note portion begins. This patch does not affect graph output or behavior in any way. Differential Revision: https://phab.mercurial-scm.org/D3655

graft: add test for reading old graftstate files with new mechanism This tests the reading of old graftstate file using the new logic. The tests shows that if user is in middle of a graft and then updates their mercurial to the version where we have new graftstate format, we can still read the old graft state format files correctly. Differential Revision: https://phab.mercurial-scm.org/D2597

graft: start using the cmdstate class to read and write data to graftstate This patch replaces the logic to read and write data to graftstate file to use the state.cmdstate() class. The previous graftstate format didn't had any version number on top of that, so we have to catch the CorruptedState error and then read the graftstate in case of old state files. This will help us to implement nice additions to graft commands like `--no-commit`, `--abort`, `--stop` flags. Passing on test-graft.t shows that things are working fine. Differential Revision: https://phab.mercurial-scm.org/D3654

graft: use state.cmdstate() to check whether graftstate exists This is a step towards make graft use the new state.cmdstate() class. This patch replaces the ugly try-except with nice if-else conditionals. Differential Revision: https://phab.mercurial-scm.org/D3652

graft: factor out function to read graft state in separate function Fatcoring out the logic in a separate function will help us in adding conditional logic for different versions of graft state files. Differential Revision: https://phab.mercurial-scm.org/D3651

state: removing remaining instances of opts class variable The cmdstate class used to have a class variable opts which used to be a dict which stored all the data for the state. Recent cleanups removed the use of that variable. There were couple of instances left which are removed by this patch. Differential Revision: https://phab.mercurial-scm.org/D3653

help: mention pattern syntax of latesttag() template function

py3: replace str.format(x) with `str % x` in githelp

githelp: drop the trailing period from single sentence output for consistency There are several instances of multiple sentence output, which I left alone. That is already nonstandard style, so dropping the period doesn't seem like an improvement.

hgweb: wrap {entries}* of filelog with mappinglist They were lists of mappings.

hgweb: don't use dict(key=value) to build a mapping dict in filelog It wasn't Py3 compatible because mapping keys must be bytes.

hgweb: wrap {diffopts} of annotate by hybriddict() The diffopts here is a plain dict, which should be wrapped by hybriddict.

hgweb: wrap {parents} of {annotate} with mappinggenerator It's a generator of at most two mappings, which has to be wrapped.

hgweb: wrap {annotate} with mappinggenerator No bare generator of mappings should be put in a template mapping.

hgweb: wrap {bookmarks} of summary with mappinggenerator No bare generator of mappings should be put in a template mapping.

hgweb: wrap {lastchange} of bookmarks with mappinglist It was an 1-length list of a mapping, can be wrapped with a mappinglist.

hgweb: wrap {entries}* of bookmarks with mappinggenerator They were functions returning a generator of mappings. The laziness is handled by the mappinggenerator class.

state: raise CorruptedState error isntead of ProgrammingError There are old state files which don't have a version number in top of them and hence we have to read them to check whether they are good or not. ProgrammingError is not apt for this case. Thanks to Yuya for suggesting CorruptedState error. Differential Revision: https://phab.mercurial-scm.org/D3644

localrepo: add docstring to _makedirstate to make it less likely to be removed Differential Revision: https://phab.mercurial-scm.org/D3643

scmutil: move repair.stripbmrevset as scmutil.bookmarkrevs (API)

githelp: cleanup one more abort message This makes the string localizable, uses the more standard hint argument, quotes the problem option so it stands out, and kills a stray apostrophe.

githelp: do not concatenate i18n messages dynamically so they can be collected

narrow: only wrap dirstate functions once, instead of per-reposetup chg will call reposetup multiple times, and we would end up double-wrapping (or worse) the dirstate functions; this can cause issues like OSError 'No such file or directory' during rebase operations, when we go to double-delete our narrowspec backup file. Differential Revision: https://phab.mercurial-scm.org/D3559

state: temporary silence pyflakes warning by removing variable assignment The variable 'version' is going to be used in upcoming series where we will be using the version number to read a certain state file. However currently, pyflakes fails because of the variable not being used. Let's remove the assignment temporarily so that buildbots and test-suite is happy until I iterate over remaining part of the series. Differential Revision: https://phab.mercurial-scm.org/D3641

state: fix usage of an unassigned variable The variable iv was used in earlier iterations of the patches and was dropped since we made sure 'version' is always int. This usage of 'iv' should have been replaced by 'version' in the original patch only. Differential Revision: https://phab.mercurial-scm.org/D3640

py3: add b'' prefixes in tests/test-revset2.t # skip-blame because just b'' prefixes Differential Revision: https://phab.mercurial-scm.org/D3637

py3: use encoding.strfromlocal() instead of pycompat.sysstr() opts['dateformat'] is provided by the user and can be encoded in local encoding. Thanks to Yuya for catching. Differential Revision: https://phab.mercurial-scm.org/D3642

tests: stabilize test-patch.t on Windows $PYTHON needs to be quoted when invoking with cmd.exe, because the value expands to c:/Python27/python.exe, which seems to be interpreted as 'c' being a command. We can't just convert to '\', because there are a few places that run $PYTHON directly in MSYS. If unquoted there, it results in c:Python27python.exe being run. I wonder if we should bake the quotes into the environment variable to avoid this. It also wasn't happy with the quoting around exit1.py: c:/Python27/python.exe: can't open file ''$TESTTMP/d/exit1.py'': [Errno 22] Invalid argument

githelp: lowercase the start of output messages for consistency I left 'Mercurial' as a proper name capitalized.