> + $ if [ -x /usr/local/bin/gmake ] ; then
> + > MAKE=gmake

Isn't this assuming that /usr/local/bin is in $PATH? IOW, shouldn't this
assignment be:

	MAKE=/usr/local/bin/gmake

Missing includes of stdint.h and stddef.h (or stdlib.h).

1. d1/f, d3/f, d3/g and d5/f should no longer be reported $ hg -R main files + not deleting possibly dirty file d3/f + not deleting possibly dirty file d3/g + not deleting possibly dirty file d5/f

+ self.repo.ui.log("rebase",
+ "using in-memory rebase: %r\n" % self.inmemory,

I couldn't find documentation on how encoding works for this (user
data interpreted by hg). This function appears to assume the
encoding of the input pattern is an extension of ascii, so I think
my change should be correct for that.

+ # Create a matcher so that prefetch doesn't attempt to fetch the entire
+ # repository pointlessly.
+ match = scmutil.matchfiles(repo, repo[node].files())

with shelvedfile(repo, name, patchextension).opener('wb') as fp:

• cmdutil.exportfile(repo, [node], fp, opts=mdiff.diffopts(git=True)) + cmdutil.exportfile(repo, [node], fp, opts=mdiff.diffopts(git=True), + match=match)

This looked a bit scary, but it should work so long as both reader and writer
sides do seek().

+ if ((next - data) < 22) {
+ /* We should have at least 22 bytes in a line:
+ 1 byte filename
+ 1 NUL
+ 20 bytes of hash

Can you add the following tests?

• with repo.wvfs(f) as fobj:
• fdata = fobj.read() + try: + with repo.wvfs(f) as fobj: + fdata = fobj.read() + except (IOError, OSError) as inst: + if inst.errno != errno.ENOENT: + raise

I didn't notice it last time, sorry.

+/* Convert a PyInt or PyLong to a long. Returns false if there is an
+ error, in which case an exception will already have been set. */
+static inline bool pylong_to_long(PyObject *pylong, long *out)

+ When multiple tools are configured to affect a file, they execute in an order
+ defined by the :priority suboption. The priority suboption has a default value
+ of zero for each tool. Tools are executed in order of ascending priority.

In this case, I suspect something really weird because `PyInt_AS_LONG()` doesn't exist on Python 3 and the docs for Python 2 say it performs no error checking. So the fact that it is setting an exception on Python 3 causes me to raise an eyebrow.

Flagged as (BC) and queued, thanks.

This name was always inaccurate, since the config accepts any pattern.
Hopefully so few people use this right now that it won't matter, but there will
now be a warning if the old config name is used.

Queued for stable, thanks.

Queued, thanks.

• a/tests/run-tests.py +++ b/tests/run-tests.py @@ -2527,13 +2527,14 @@ os.umask(oldmask)
  
  def _run(self, testdescs): + testdir = getcwdb()

if __name__ == '__main__':

+ if (3, 6, 0) <= sys.version_info <= (3, 6, 3):
+ # Python 3.6.0 through 3.6.3 inclusive shipped with
+ # https://bugs.python.org/issue31825 and we can't run these
+ # tests on those specific versions of Python. Sigh.
+ sys.exit(0)

Looks good, but test-check-config.t failed probably because the doc couldn't
be found by the checker.

Maybe missing b'' ?

>   class mappingdict(mappable, _mappingsequence):
>       """Wrapper for a single template mapping
>   
>       This isn't a sequence in a way that the underlying dict won't be iterated
>       as a dict, but shares most of the _mappingsequence functions.
>       """
>   
>       def __init__(self, mapping, name=None, tmpl=None):
>           super(mappingdict, self).__init__(name, tmpl)
>           self._mapping = mapping
>   
>       def tomap(self, context):
>           return self._mapping
>   
>       def tobool(self, context, mapping):
>           # no idea when a template mapping should be considered an empty, but
>           # a mapping dict should have at least one item in practice, so always
>           # mark this as non-empty.
>           return True
>   
>       def tovalue(self, context, mapping):
>           return super(mappingdict, self).tovalue(context, mapping)[0]

I copy/pasted this to https://phab.mercurial-scm.org/D5211. :)

> Apparently thg uses this to avoid loading the data into memory just to figure out the length like fctx.rawsize().  Any suggestions?
>
> https://groups.google.com/d/msg/thg-dev/6ekYID7ho_o/fbdd6rVDAwAJ


Just found this comment when going through mail backlog.

Do we want to do anything about this before the 4.8 release?

> >   We still need to linearize the revisions in some cases, introduce an explicit
> >   `linear` sorting before changing back the default order.
>
> I'm a little confused about this. We do introduce `nodesorder='linear'`, but
>  there seems no user of that option. Are there unsent patches?


The message should have said `We **may** still need to`. There is no user of that option yet so we can drop it.

I think we should at least keep the option in `storageutil.emitrevision` so we can experiment around doing a DAG sort for various operations, @indygreg seems to have some perf improvements on the firefox repo with it

We still need to linearize the revisions in some cases, introduce an explicit
`linear` sorting before changing back the default order.

Looks good and simple. Queued for stable, thanks.

Nit: ui as uimod per test-check-code.t.

• a/mercurial/help.py +++ b/mercurial/help.py @@ -391,6 +391,12 @@ single documentation block and use it to overwrite the marker in doc. """ entries = [] + markerindent = 0 + markerpos = doc.find(marker) + if markerpos >= 0: + newlinepos = doc.rfind('\n', 0, markerpos) + if newlinepos >= 0: + markerindent = markerpos - newlinepos - 1

Queued the series for 4.9, many thanks.

+ # extract the canonical command from the passed opts.
+ # we need to pop it up from the opts else the msg will try
+ # being formatted using the passed opts.

Do we have performance concerns around the deprecation, how often is called `loadcmdtable`? My gut feeling is telling me it shouldn't matter but I would prefer to be sure.

FYI, I've posted a similar fix a few days ago.

> rbc = self.revbranchcache()
> + cl = self.changelog

Perhaps it has to be self.unfiltered().changelog because revbranchcache is
initialized with the unfiltered repo.