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.