Mon, 10 Jul 2017 19:08:17 +0200 bookmark: split out target computation from 'deletedivergent'
Boris Feld <boris.feld@octobus.net> [Mon, 10 Jul 2017 19:08:17 +0200] rev 33510
bookmark: split out target computation from 'deletedivergent' We want to use applychanges in order to unify bookmark movement. We need a way to compute divergence deletion without actually removing them. We split the function in two in this patch while we migrate the existing users of this code on next patches.
Mon, 10 Jul 2017 17:48:33 +0200 bookmark: remove a useless 'recordchange' in the amend code
Boris Feld <boris.feld@octobus.net> [Mon, 10 Jul 2017 17:48:33 +0200] rev 33509
bookmark: remove a useless 'recordchange' in the amend code We do not touch the bookmarks store in this code, just the active bookmark, not covered by the transaction. So it seems we can safely drop this call and the tests agree with us.
Sat, 15 Jul 2017 23:46:33 -0400 test-phase-exchange: stabilize for Windows
Matt Harbison <matt_harbison@yahoo.com> [Sat, 15 Jul 2017 23:46:33 -0400] rev 33508
test-phase-exchange: stabilize for Windows
Sat, 15 Jul 2017 15:23:29 -0400 debugignore: eliminate inconsistencies with `hg status` (issue5222)
Matt Harbison <matt_harbison@yahoo.com> [Sat, 15 Jul 2017 15:23:29 -0400] rev 33507
debugignore: eliminate inconsistencies with `hg status` (issue5222) Using a matcher for this command allows processing the named file(s) as relative to cwd. It also leverages the icasefs normalization logic the same way the status command does. (However, a false indicator is given for a nonexistent file in some cases, e.g. passing 'foo.REJ' when that file doesn't exist, and the rule is '*.rej'. Maybe the regex itself needs to be case insensitive on these platforms, at least for the debug command.) Finally, the file printed is relative to cwd and uses platform specific slashes, so a few (glob)s were needed in seemingly unrelated tests.
Sun, 16 Jul 2017 04:39:32 -0700 commandserver: close selector explicitly
Jun Wu <quark@fb.com> [Sun, 16 Jul 2017 04:39:32 -0700] rev 33506
commandserver: close selector explicitly The selector does not have a __del__ method and needs a manual close. We can also use "with selector" but that makes the code too indented. Therefore append a "selector.close()" after the end of the main loop for now.
Sat, 15 Jul 2017 15:01:29 +0900 scmutil: remove duplicated import of i18n._()
Yuya Nishihara <yuya@tcha.org> [Sat, 15 Jul 2017 15:01:29 +0900] rev 33505
scmutil: remove duplicated import of i18n._()
Sun, 04 Jun 2017 10:02:09 -0700 obsstore: let read marker API take a range of offsets
Jun Wu <quark@fb.com> [Sun, 04 Jun 2017 10:02:09 -0700] rev 33504
obsstore: let read marker API take a range of offsets This allows us to read a customized range of markers, instead of loading all of them. The condition of stop is made consistent across C and Python implementation so we will still read marker when offset=a, stop=a+1.
Fri, 14 Jul 2017 20:26:21 -0700 commandserver: use selectors2
Jun Wu <quark@fb.com> [Fri, 14 Jul 2017 20:26:21 -0700] rev 33503
commandserver: use selectors2 Previously, commandserver was using select.select. That could have issue if _sock.fileno() >= FD_SETSIZE (usually 1024), which raises: ValueError: filedescriptor out of range in select() We got that in production today, although it's the code opening that many files to blame, it seems better for commandserver to work in this case. There are multiple way to "solve" it, like preserving a fd with a small number and swap it with sock using dup2(). But upgrading to a modern selector supported by the system seems to be the most correct way.
(0) -30000 -10000 -3000 -1000 -300 -100 -30 -10 -8 +8 +10 +30 +100 +300 +1000 +3000 +10000 tip