##// END OF EJS Templates
discovery: slowly increase sampling size...
discovery: slowly increase sampling size Some pathological discovery runs can requires many roundtrip. When this happens things can get very slow. To make the algorithm more resilience again such pathological case. We slowly increase the sample size with each roundtrip (+5%). This will have a negligible impact on "normal" discovery with few roundtrips, but a large positive impact of case with many roundtrips. Asking more question per roundtrip helps to reduce the undecided set faster. Instead of reducing the undecided set a linear speed (in the worst case), we reduce it as a guaranteed (small) exponential rate. The data below show this slow ramp up in sample size: round trip | 1 | 5 | 10 | 20 | 50 | 100 | 130 | sample size | 200 | 254 | 321 | 517 | 2 199 | 25 123 | 108 549 | covered nodes | 200 | 1 357 | 2 821 | 7 031 | 42 658 | 524 530 | 2 276 755 | To be a bit more concrete, lets take a very pathological case as an example. We are doing discovery from a copy of Mozilla-try to a more recent version of mozilla-unified. Mozilla-unified heads are unknown to the mozilla-try repo and there are over 1 million "missing" changesets. (the discovery is "local" to avoid network interference) Without this change, the discovery: - last 1858 seconds (31 minutes), - does 1700 round trip, - asking about 340 000 nodes. With this change, the discovery: - last 218 seconds (3 minutes, 38 seconds a -88% improvement), - does 94 round trip (-94%), - asking about 344 211 nodes (+1%). Of course, this is an extreme case (and 3 minutes is still slow). However this give a good example of how this sample size increase act as a safety net catching any bad situations. We could image a steeper increase than 5%. For example 10% would give the following number: round trip | 1 | 5 | 10 | 20 | 50 | 75 | 100 | sample size | 200 | 321 | 514 | 1 326 | 23 060 | 249 812 | 2 706 594 | covered nodes | 200 | 1 541 | 3 690 | 12 671 | 251 871 | 2 746 254 | 29 770 966 | In parallel, it is useful to understand these pathological cases and improve them. However the current change provides a general purpose safety net to smooth the impact of pathological cases. To avoid issue with older http server, the increase in sample size only occurs if the protocol has not limit on command argument size.

File last commit:

r38915:e79a69af default
r42546:dbd0fcca default
Show More
minifileset.py
92 lines | 3.5 KiB | text/x-python | PythonLexer
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 # minifileset.py - a simple language to select files
#
# Copyright 2017 Facebook, Inc.
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
from __future__ import absolute_import
from .i18n import _
from . import (
error,
fileset,
Yuya Nishihara
fileset: extract language processing part to new module (API)...
r38841 filesetlang,
Augie Fackler
minifileset: fix on Python 3...
r37893 pycompat,
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 )
Yuya Nishihara
fileset: parse argument of size() by predicate function...
r38709 def _sizep(x):
# i18n: "size" is a keyword
Yuya Nishihara
fileset: extract language processing part to new module (API)...
r38841 expr = filesetlang.getstring(x, _("size requires an expression"))
Yuya Nishihara
fileset: parse argument of size() by predicate function...
r38709 return fileset.sizematcher(expr)
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 def _compile(tree):
if not tree:
raise error.ParseError(_("missing argument"))
op = tree[0]
Yuya Nishihara
fileset: insert hints where status should be computed...
r38915 if op == 'withstatus':
return _compile(tree[1])
elif op in {'symbol', 'string', 'kindpat'}:
Yuya Nishihara
fileset: extract language processing part to new module (API)...
r38841 name = filesetlang.getpattern(tree, {'path'}, _('invalid file pattern'))
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 if name.startswith('**'): # file extension test, ex. "**.tar.gz"
ext = name[2:]
Augie Fackler
minifileset: fix on Python 3...
r37893 for c in pycompat.bytestr(ext):
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 if c in '*{}[]?/\\':
raise error.ParseError(_('reserved character: %s') % c)
return lambda n, s: n.endswith(ext)
Yuya Nishihara
minifileset: unify handling of symbol and string patterns...
r35758 elif name.startswith('path:'): # directory or full path test
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 p = name[5:] # prefix
pl = len(p)
Augie Fackler
minifileset: fix on Python 3...
r37893 f = lambda n, s: n.startswith(p) and (len(n) == pl
or n[pl:pl + 1] == '/')
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 return f
Matt Harbison
minifileset: note the unsupported file pattern when raising a parse error...
r35818 raise error.ParseError(_("unsupported file pattern: %s") % name,
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 hint=_('paths must be prefixed with "path:"'))
Yuya Nishihara
fileset: combine union of basic patterns into single matcher...
r38901 elif op in {'or', 'patterns'}:
Yuya Nishihara
fileset: flatten 'or' nodes to unnest unionmatchers...
r38840 funcs = [_compile(x) for x in tree[1:]]
return lambda n, s: any(f(n, s) for f in funcs)
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 elif op == 'and':
func1 = _compile(tree[1])
func2 = _compile(tree[2])
return lambda n, s: func1(n, s) and func2(n, s)
elif op == 'not':
return lambda n, s: not _compile(tree[1])(n, s)
elif op == 'func':
symbols = {
'all': lambda n, s: True,
'none': lambda n, s: False,
Yuya Nishihara
fileset: parse argument of size() by predicate function...
r38709 'size': lambda n, s: _sizep(tree[2])(s),
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 }
Yuya Nishihara
fileset: extract language processing part to new module (API)...
r38841 name = filesetlang.getsymbol(tree[1])
Yuya Nishihara
fileset: make it robust for bad function calls...
r35709 if name in symbols:
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 return symbols[name]
raise error.UnknownIdentifier(name, symbols.keys())
elif op == 'minus': # equivalent to 'x and not y'
func1 = _compile(tree[1])
func2 = _compile(tree[2])
return lambda n, s: func1(n, s) and not func2(n, s)
elif op == 'list':
raise error.ParseError(_("can't use a list in this context"),
Martin von Zweigbergk
help: add quotes to a few commands we point to...
r38846 hint=_('see \'hg help "filesets.x or y"\''))
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 raise error.ProgrammingError('illegal tree: %r' % (tree,))
def compile(text):
"""generate a function (path, size) -> bool from filter specification.
"text" could contain the operators defined by the fileset language for
common logic operations, and parenthesis for grouping. The supported path
tests are '**.extname' for file extension test, and '"path:dir/subdir"'
for prefix test. The ``size()`` predicate is borrowed from filesets to test
file size. The predicates ``all()`` and ``none()`` are also supported.
Yuya Nishihara
fileset: add kind:pat operator...
r35759 '(**.php & size(">10MB")) | **.zip | (path:bin & !path:bin/README)' for
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 example, will catch all php files whose size is greater than 10 MB, all
files whose name ends with ".zip", and all files under "bin" in the repo
root except for "bin/README".
"""
Yuya Nishihara
fileset: extract language processing part to new module (API)...
r38841 tree = filesetlang.parse(text)
Yuya Nishihara
fileset: add phase to transform parsed tree...
r38862 tree = filesetlang.analyze(tree)
Yuya Nishihara
fileset: add stub for weight-based optimization...
r38865 tree = filesetlang.optimize(tree)
Matt Harbison
fileset: add a lightweight file filtering language...
r35634 return _compile(tree)