patch: fix import-time syntax error in test-check-module-imports.t This is more correct, though we were getting lucky in practice with our rewriter saving us. # skip-blame just a b prefix Differential Revision: https://phab.mercurial-scm.org/D3584

py3: whitelist two more passing tests observed by buildbot Differential Revision: https://phab.mercurial-scm.org/D3583

py3: use stringutil.pprint() to format a list to print Differential Revision: https://phab.mercurial-scm.org/D3582

py3: fix kwargs handling in qgurad in hgext/mq.py # skip-blame because just r'' prefixes Differential Revision: https://phab.mercurial-scm.org/D3581

py3: use .startswith() instead of bytes[0] bytes[0] returns the ascii value of character at 0 index. Differential Revision: https://phab.mercurial-scm.org/D3580

crecord: fallback to text mode if diffs are too big for curses mode crecord uses curses.newpad to create a region that we can then scroll around in by moving the main 'screen' as a veiwport into the (probably larger than the actual screen) pad. Internally, at least in ncurses, pads are implemented using windows, which have their dimensions limited to a certain size. Depending on compilation options for ncurses, this size might be pretty small: (signed) short, or it might be larger ((signed) int). crecord wants to have enough room to have all of the contents of the main area of the chunkselector in the pad; this means that the full size with everything expanded must be less than these (undocumented, afaict) limits. It's not easy to write tests for this because the limits are platform- and installation- dependent and undocumented / unqueryable, as far as I can tell. Differential Revision: https://phab.mercurial-scm.org/D3577

narrow: filter copies in core Differential Revision: https://phab.mercurial-scm.org/D3576

narrow: filter set of files to check for case-folding to core Differential Revision: https://phab.mercurial-scm.org/D3575

narrow: filter merge actions in core Part of the ongoing effort to move narrow into core. Differential Revision: https://phab.mercurial-scm.org/D3574

hgweb: extract code for emitting multiple changelist records changelistentry() exists so extensions can easily monkeypatch the function to add additional metadata. In at least one case at Mozilla, we have an extension doing this where performance would greatly benefit if we were able to "batch fetch" metadata for all revisions that will eventually have their data set. By extracting the logic for "obtain records for N revisions" into a standalone function, we make it possible for a monkeypatched function to more efficiently bulk fetch data for N revisions. Differential Revision: https://phab.mercurial-scm.org/D3573

phabricator: register config settings I didn't bother registering the deprecated phabricator.auth.*, and I'm not sure if the two flagged as 'developer config' should be moved to [devel] (or why there's a distinction between `repophid` and `callsign`).

filemerge: support specifying a python function to custom merge-tools Eliminates the need to specify a python executable, which may not exist on system. Additionally launching script inprocess aids portability on systems that can't execute python via the shell. Example usage "merge-tools.myTool.executable=python:c:\myTool.py:mergefn" where myTool.py contains a function: "def mergefn(ui, repo, args, **kwargs):" where args is list of args passed to merge tool. (by default, expanded: $local $base $other) Invoking the specified python function was done by exposing and invoking (hook._pythonhook -> hook.pythonhook)

help: correct signature of separate() template function Without the dots, it looked as if separate() would take a list of arguments.

hgweb: do not try to replace signal handlers while locking According to the issue 5889, mod_wsgi issues a warning on signal.signal() call, and we wouldn't want to see it in error log. The problem addressed by d77c3b023393 could potentially occur in web session, but that would be less likely than in user processes.

lock: add internal config to not replace signal handlers while locking signal.signal() is blocked in some WSGI environments, and a horrible warning is sent to the server log. So we need a way to disable it, and I think abusing ui.config is the simplest workaround.

merge with i18n