convert: use raw strings for XML strings Due to the source transformer, we were passing bytes into the XML APIs. This results in not finding elements and doing compares against mismatched types. Use raw string literals so we use str everywhere. Differential Revision: https://phab.mercurial-scm.org/D5664

tests: normalize XML values to bytes This fixes some type mismatches. The encoding shouldn't matter for what this script is used for. UTF-8 seems reasonable, especially since I'm pretty sure SVN will emit UTF-8 encoded XML. Differential Revision: https://phab.mercurial-scm.org/D5665

hgweb: ensure template mapping keys are bytes Before, str keys were being added in Python 3 because named arguments to dict() use native str for keys. This caused the templater to fail to find the keys since it was looking for bytes versions. This makes a handful of tests pass on Python 3. We may want to consider having the templater validate that keys in mapping dicts are bytes. But I'm unsure whether this is appropriate and won't be doing this. Differential Revision: https://phab.mercurial-scm.org/D5666

remotefilelog: use %d to format an int Differential Revision: https://phab.mercurial-scm.org/D5656

tests: use bytes for file I/O Otherwise we get various type mismatches. Differential Revision: https://phab.mercurial-scm.org/D5655

tests: make filenames bytes for Python 3 I also snuck a %s -> %d in there to appease Python 3. Differential Revision: https://phab.mercurial-scm.org/D5654

tests: use assertEqual() This avoids a deprecation warning under at least Python 3.7. Differential Revision: https://phab.mercurial-scm.org/D5653

tests: add setsockopt() output on Python 3 Python 3 appears to call setsockopt() where Python 2 did not. Differential Revision: https://phab.mercurial-scm.org/D5652

hgweb: don't use raw string for session vars This r'' is leaking into the templating layer and causing an assertion failure. The r'' was added in d1fccbd50fcd (October 2017). Similar code in hgweb_mod.py was also changed in that changeset. hgweb_mod.py was updated in ec46415ed826 (March 2018) to use webutil.sessionvars(), which doesn't use raw strings. Differential Revision: https://phab.mercurial-scm.org/D5651

tests: normalize to bytes in test-install.t directory() was returning str and hgdirectory() was returning bytes. This made the set compare fail. Let's normalize the types on Python 3 so the test passes. Differential Revision: https://phab.mercurial-scm.org/D5650

config: reject str sections and keys on Python 3 Otherwise we could end up with a dict having both the str and bytes versions of a particular config item. This may cause some tests to regress. I haven't checked. But it is better behavior to fail fast. We could just as easily change this to normalize the values. But I like catching all non-compliant call sites first. Differential Revision: https://phab.mercurial-scm.org/D5649

tests: add size and hash for Python 3 test-archive.t now passes on Python 3.7.1. Differential Revision: https://phab.mercurial-scm.org/D5648

httppeer: use %s for formatting This prevents a b'' from appearing in Python 3 output. We keep the single quotes for backwards compatibility. Differential Revision: https://phab.mercurial-scm.org/D5647

tests: add b'' to notcapable This fixes some exceptions in a few tests. # skip-blame just b'' prefixes Differential Revision: https://phab.mercurial-scm.org/D5646

tests: add b'' to test-missing-capability.t The test now passes on Python 3. # skip-blame just b'' prefixes Differential Revision: https://phab.mercurial-scm.org/D5645

fuzz: exercise more of the lazymanifest code Differential Revision: https://phab.mercurial-scm.org/D5643

fuzz: restrict manifest input size Again, let's keep the fuzzer from getting excited about huge inputs. Differential Revision: https://phab.mercurial-scm.org/D5642

fuzz: exercise more of the revlog API I noticed in the coverage report that we didn't have much coverage in revlog.py. Let's try and get some of the more interesting bits tested by the fuzzer. I ran this locally for a few minutes to verify that I appear to be calling the various functions in reasonable ways. Differential Revision: https://phab.mercurial-scm.org/D5641

fuzz: don't allow enormous revlog inputs either I'm about to make the fuzzer do more, and without this it was getting enthusiastic about large (and therefore slow) inputs that I don't think buy us much. Differential Revision: https://phab.mercurial-scm.org/D5640

ui: remove unreachable branches and function calls from write() (issue6059) This is at least faster than ui.write() of 4.8.2. $ HGRCPATH=/dev/null hg files -R mozilla-central --time >/dev/null 4.8.2: time: real 2.340 secs (user 2.310+0.000 sys 0.020+0.000) 4.9rc0: time: real 2.580 secs (user 2.550+0.000 sys 0.020+0.000) this: time: real 2.230 secs (user 2.210+0.000 sys 0.020+0.000) Maybe the formatter should own a resolved write() function because it will just call dest.write(msg) most of the time, but that would be too much for stable.

ui: inline _writenobuf() into write() due to performance issue I'll remove redundant conditions later in this series.

ui: inline _write() into write() due to performance issue I'll remove redundant conditions later in this series.

ui: optimize buffered write with no label This was spotted while making fastannotate faster again after ditching its own formatter. Since I'm going to inline _write() into ui.write(), I decided to include this patch in this series. Here, the cost of '(self.label(a, label) for a in args)' was significant in hot loops.

partialdiscovery: avoid `undecided` related computation sooner than necessary Changeset 1d30be90c move the update of the `undecided` set within the `partialdiscovery` object in order to clarify the API. The update to the `undecided` set was unconditional in 1d30be90c and the first access to the `self.undecided` property triggered the initial computation of the set of undecided revisions. As a result, the set was computed much earlier, at a time where less information is available, immediately followed by an update of this set to remove common revisions. To fix this regression, we ignore the `undecided` related logic in `addcommons` when that `undecided` set has not been computed yet. Code that actually needs to know the `undecided` set will trigger its computation later. The change has no effects on semantic because the initial computation `undecided` set takes all knowns `common` into account. Example performance running `hg debugdiscovery` from a pypy repo missing 10 changesets: 870a89c6909d: 52.3ms (regression parent) 1d30be90c9dc: 72.0ms (regression) 5a5f504a7175: 64.8ms (this fix parent) this fix: 52.6ms

revlog: fix resolution of revlog version 0 This partially backs out cecf3f8bccd3, "revlog: always process opener options." My understanding is that if there's no "revlog1" nor "revlog2" in .hg/requires, the repository should stick to the v0 format. The reasoning is briefly described in 31a5973fcf96, "revlog: get rid of defversion." Maybe we can drop support for missing opener options, but I didn't do that in this patch.

merge with stable

rust: add comment about lack of wdirrev handling If hg is compiled with rust support, 'only(wdir())' crashed as "rustext.GraphError: ('ParentOutOfRange', 2147483647)", which should instead say "abort: working directory revision cannot be specified."

templatekw: fix crash on multiple latesttags resolution at wdir (issue6055) It appears not easy to fix only() to support wdir(), so this patch works around the issue by getlatesttags(). The "+1" after len(changes) doesn't matter since ctx never changes while sorting. It's just for clarity.

test-template-keywords: add test for {latesttag} of wdir() revision It's probably broken since fb672eac2702, "templatekw: choose {latesttag} by len(changes), not date (issue5659)". only() doesn't support wdir.

narrow: fix crash when restoring backup in legacy repo Using --addremove when committing in an old repo (before we started keeping .hg/narrowspec.dirstate) results in a crash. The test case modified in this patch would crash like this: abort: $ENOENT$ The issue is that when the dirstateguard is aborted, it tries to restore the backup of .hg/narrowspec.dirstate. However, since we were in an old repo, that file did not get created when the dirstateguard was created. Note that the dirstateguard is not used unless --addremove is passed. This patch fixes the bug by making restorewcbackup() not fail if the backup doesn't exist. I also made clearwcbackup() safe, just in case. Differential Revision: https://phab.mercurial-scm.org/D5634