test-amend: match output using conditional test case name D466 (6cc8f848b4c3) allows output to be conditionally matched by test name. This patch changes test-amend.t to use that feature, instead of duplicating `hg amend` command or use `-q` to silence its output. Differential Revision: https://phab.mercurial-scm.org/D601

util: use set for reserved Windows filenames Previously, we were performing membership testing against a list. Change it to a set for a minor perf win. While we're at it, explode the assignment in place so less work is needed at module import time. Differential Revision: https://phab.mercurial-scm.org/D600

context: add arbitraryfilectx, which can represent files outside the workdir Move it from contrib/simplemerge so it can be re-used in the future. Differential Revision: https://phab.mercurial-scm.org/D604

simplemerge: remove unused `filtereddata` parameter Differential Revision: https://phab.mercurial-scm.org/D603

simplemerge: remove unused `repo` parameter This is now no longer used or needed thanks to the `decodeddata()` context function. Differential Revision: https://phab.mercurial-scm.org/D602

extensions: prohibit unicode defaults If the default value of an option is a unicode string (something than happen easily when using a 'from __future__ import unicode_literals'), any value passed on the command line will be ignored because the fancyopts module only checks for byte strings and not unicode strings. Changing fancyopts behavior is easy but would make assumptions on how the python3 port should be done, which is outside the scope of this patch. The chosen approach is to stop an extension from being loaded when a unicode default value is detected, with a hint for the developer.

revsetlang: remove unused functions Superseded by the _match() function.

revsetlang: match tree by helper function on optimize This should make optimize() more readable and less error-prone, but it doubles the parsing cost. (original) $ python -m timeit -n10000 -s 'from mercurial import revsetlang as L' \ 'L.optimize(L.analyze(L.parse("ancestors(x) and not ancestors(y)")))' 10000 loops, best of 3: 79.3 usec per loop (this patch) $ python -m timeit -n10000 -s 'from mercurial import revsetlang as L' \ 'L._treecache.clear(); \ L.optimize(L.analyze(L.parse("ancestors(x) and not ancestors(y)")))' 10000 loops, best of 3: 201 usec per loop

parser: add helper function to test if pattern matches parsed tree This function will be used as follows: match('ancestors(_) and not ancestors(_)', x) See the next patch for details.

revsetlang: build optimized tree by helper function This should make optimize() more readable, but it doubles the parsing cost. (original) $ python -m timeit -n10000 -s 'from mercurial import revsetlang as L' \ 'L.optimize(L.analyze(L.parse("::tip")))' 10000 loops, best of 3: 18.1 usec per loop (this patch) $ python -m timeit -n10000 -s 'from mercurial import revsetlang as L' \ 'L._treecache.clear(); L.optimize(L.analyze(L.parse("::tip")))' 10000 loops, best of 3: 48.4 usec per loop 30usec isn't dominant compared to the revset evaluation, but that is a cost. That's why a parsed tree is cached, which can benefit in hgweb or chg server.

parser: add helper function that constructs parsed tree from template This function will be used as follows: build('only(_, _)', x, y) See the next patch for details.

patch: take messages out of the function so that extensions can add entries Extensions will want to have interactive thing for more operations or particulary want to show more verbs. So this patch takes out the message thing from the function so that extensions can add verbs to this. The curses one is also not in any function so extensions can add more actions and verbs there. Differential Revision: https://phab.mercurial-scm.org/D567