rust: peek_mut optim for lazy ancestors
This is one of the two optimizations that are also
present in the Python code: replacing pairs of pop/push
on the BinaryHeap by single updates, hence having it
under the hood maintain its consistency (sift) only once.
On Mozilla central, the measured gain (see details below)
is around 7%.
Creating the PeekMut object by calling peek_mut() right away
instead of peek() first is less efficient (gain is only 4%, stats
not included).
Our interpretation is that its creation has a cost which is vasted
in the cases where it ends by droping the value (Peekmut::pop()
just does self.heap.pop() anyway). On the other hand, the immutable
peek() is very fast: it's just taking a reference in the
underlying vector.
The Python version still has another optimization:
if parent(current) == current-1, then the heap doesn't need
to maintain its consistency, since we already know that
it's bigger than all the others in the heap.
Rust's BinaryHeap doesn't allow us to mutate its biggest
element with no housekeeping, but we tried it anyway, with a
copy of the BinaryHeap implementation with a dedicaded added
method: it's not worth the technical debt in our opinion
(we measured only a further 1.6% improvement).
One possible explanation would be that the sift is really fast
anyway in that case, whereas it's not in the case of Python,
because it's at least partly done in slow Python code.
Still it's possible that replacing BinaryHeap by something more
dedicated to discrete ordered types could be faster.
Measurements on mozilla-central:
Three runs of 'hg perfancestors' on the parent changeset:
Moyenne des médianes: 0.100587
! wall 0.100062 comb 0.100000 user 0.100000 sys 0.000000 (best of 98)
! wall 0.135804 comb 0.130000 user 0.130000 sys 0.000000 (max of 98)
! wall 0.102864 comb 0.102755 user 0.099286 sys 0.003469 (avg of 98)
! wall 0.101486 comb 0.110000 user 0.110000 sys 0.000000 (median of 98)
! wall 0.096804 comb 0.090000 user 0.090000 sys 0.000000 (best of 100)
! wall 0.132235 comb 0.130000 user 0.120000 sys 0.010000 (max of 100)
! wall 0.100258 comb 0.100300 user 0.096000 sys 0.004300 (avg of 100)
! wall 0.098384 comb 0.100000 user 0.100000 sys 0.000000 (median of 100)
! wall 0.099925 comb 0.100000 user 0.100000 sys 0.000000 (best of 98)
! wall 0.133518 comb 0.140000 user 0.130000 sys 0.010000 (max of 98)
! wall 0.102381 comb 0.102449 user 0.098265 sys 0.004184 (avg of 98)
! wall 0.101891 comb 0.090000 user 0.090000 sys 0.000000 (median of 98)
Mean of the medians: 0.100587
On the present changeset:
! wall 0.091344 comb 0.090000 user 0.090000 sys 0.000000 (best of 100)
! wall 0.122728 comb 0.120000 user 0.110000 sys 0.010000 (max of 100)
! wall 0.093268 comb 0.093300 user 0.089300 sys 0.004000 (avg of 100)
! wall 0.092567 comb 0.100000 user 0.090000 sys 0.010000 (median of 100)
! wall 0.093294 comb 0.080000 user 0.080000 sys 0.000000 (best of 100)
! wall 0.144887 comb 0.150000 user 0.140000 sys 0.010000 (max of 100)
! wall 0.097708 comb 0.097700 user 0.093400 sys 0.004300 (avg of 100)
! wall 0.094980 comb 0.100000 user 0.090000 sys 0.010000 (median of 100)
! wall 0.091262 comb 0.090000 user 0.080000 sys 0.010000 (best of 100)
! wall 0.123772 comb 0.130000 user 0.120000 sys 0.010000 (max of 100)
! wall 0.093188 comb 0.093200 user 0.089300 sys 0.003900 (avg of 100)
! wall 0.092364 comb 0.100000 user 0.090000 sys 0.010000 (median of 100)
Mean of the medians is 0.0933
Differential Revision: https://phab.mercurial-scm.org/D5358
from __future__ import absolute_import
import binascii
import itertools
import silenttestrunner
import unittest
import zlib
from mercurial import (
manifest as manifestmod,
match as matchmod,
)
EMTPY_MANIFEST = b''
HASH_1 = b'1' * 40
BIN_HASH_1 = binascii.unhexlify(HASH_1)
HASH_2 = b'f' * 40
BIN_HASH_2 = binascii.unhexlify(HASH_2)
HASH_3 = b'1234567890abcdef0987654321deadbeef0fcafe'
BIN_HASH_3 = binascii.unhexlify(HASH_3)
A_SHORT_MANIFEST = (
b'bar/baz/qux.py\0%(hash2)s%(flag2)s\n'
b'foo\0%(hash1)s%(flag1)s\n'
) % {b'hash1': HASH_1,
b'flag1': b'',
b'hash2': HASH_2,
b'flag2': b'l',
}
A_DEEPER_MANIFEST = (
b'a/b/c/bar.py\0%(hash3)s%(flag1)s\n'
b'a/b/c/bar.txt\0%(hash1)s%(flag1)s\n'
b'a/b/c/foo.py\0%(hash3)s%(flag1)s\n'
b'a/b/c/foo.txt\0%(hash2)s%(flag2)s\n'
b'a/b/d/baz.py\0%(hash3)s%(flag1)s\n'
b'a/b/d/qux.py\0%(hash1)s%(flag2)s\n'
b'a/b/d/ten.txt\0%(hash3)s%(flag2)s\n'
b'a/b/dog.py\0%(hash3)s%(flag1)s\n'
b'a/b/fish.py\0%(hash2)s%(flag1)s\n'
b'a/c/london.py\0%(hash3)s%(flag2)s\n'
b'a/c/paper.txt\0%(hash2)s%(flag2)s\n'
b'a/c/paris.py\0%(hash2)s%(flag1)s\n'
b'a/d/apple.py\0%(hash3)s%(flag1)s\n'
b'a/d/pizza.py\0%(hash3)s%(flag2)s\n'
b'a/green.py\0%(hash1)s%(flag2)s\n'
b'a/purple.py\0%(hash2)s%(flag1)s\n'
b'app.py\0%(hash3)s%(flag1)s\n'
b'readme.txt\0%(hash2)s%(flag1)s\n'
) % {b'hash1': HASH_1,
b'flag1': b'',
b'hash2': HASH_2,
b'flag2': b'l',
b'hash3': HASH_3,
}
HUGE_MANIFEST_ENTRIES = 200001
izip = getattr(itertools, 'izip', zip)
if 'xrange' not in globals():
xrange = range
A_HUGE_MANIFEST = b''.join(sorted(
b'file%d\0%s%s\n' % (i, h, f) for i, h, f in
izip(xrange(200001),
itertools.cycle((HASH_1, HASH_2)),
itertools.cycle((b'', b'x', b'l')))))
class basemanifesttests(object):
def parsemanifest(self, text):
raise NotImplementedError('parsemanifest not implemented by test case')
def testEmptyManifest(self):
m = self.parsemanifest(EMTPY_MANIFEST)
self.assertEqual(0, len(m))
self.assertEqual([], list(m))
def testManifest(self):
m = self.parsemanifest(A_SHORT_MANIFEST)
self.assertEqual([b'bar/baz/qux.py', b'foo'], list(m))
self.assertEqual(BIN_HASH_2, m[b'bar/baz/qux.py'])
self.assertEqual(b'l', m.flags(b'bar/baz/qux.py'))
self.assertEqual(BIN_HASH_1, m[b'foo'])
self.assertEqual(b'', m.flags(b'foo'))
with self.assertRaises(KeyError):
m[b'wat']
def testSetItem(self):
want = BIN_HASH_1
m = self.parsemanifest(EMTPY_MANIFEST)
m[b'a'] = want
self.assertIn(b'a', m)
self.assertEqual(want, m[b'a'])
self.assertEqual(b'a\0' + HASH_1 + b'\n', m.text())
m = self.parsemanifest(A_SHORT_MANIFEST)
m[b'a'] = want
self.assertEqual(want, m[b'a'])
self.assertEqual(b'a\0' + HASH_1 + b'\n' + A_SHORT_MANIFEST,
m.text())
def testSetFlag(self):
want = b'x'
m = self.parsemanifest(EMTPY_MANIFEST)
# first add a file; a file-less flag makes no sense
m[b'a'] = BIN_HASH_1
m.setflag(b'a', want)
self.assertEqual(want, m.flags(b'a'))
self.assertEqual(b'a\0' + HASH_1 + want + b'\n', m.text())
m = self.parsemanifest(A_SHORT_MANIFEST)
# first add a file; a file-less flag makes no sense
m[b'a'] = BIN_HASH_1
m.setflag(b'a', want)
self.assertEqual(want, m.flags(b'a'))
self.assertEqual(b'a\0' + HASH_1 + want + b'\n' + A_SHORT_MANIFEST,
m.text())
def testCopy(self):
m = self.parsemanifest(A_SHORT_MANIFEST)
m[b'a'] = BIN_HASH_1
m2 = m.copy()
del m
del m2 # make sure we don't double free() anything
def testCompaction(self):
unhex = binascii.unhexlify
h1, h2 = unhex(HASH_1), unhex(HASH_2)
m = self.parsemanifest(A_SHORT_MANIFEST)
m[b'alpha'] = h1
m[b'beta'] = h2
del m[b'foo']
want = b'alpha\0%s\nbar/baz/qux.py\0%sl\nbeta\0%s\n' % (
HASH_1, HASH_2, HASH_2)
self.assertEqual(want, m.text())
self.assertEqual(3, len(m))
self.assertEqual([b'alpha', b'bar/baz/qux.py', b'beta'], list(m))
self.assertEqual(h1, m[b'alpha'])
self.assertEqual(h2, m[b'bar/baz/qux.py'])
self.assertEqual(h2, m[b'beta'])
self.assertEqual(b'', m.flags(b'alpha'))
self.assertEqual(b'l', m.flags(b'bar/baz/qux.py'))
self.assertEqual(b'', m.flags(b'beta'))
with self.assertRaises(KeyError):
m[b'foo']
def testSetGetNodeSuffix(self):
clean = self.parsemanifest(A_SHORT_MANIFEST)
m = self.parsemanifest(A_SHORT_MANIFEST)
h = m[b'foo']
f = m.flags(b'foo')
want = h + b'a'
# Merge code wants to set 21-byte fake hashes at times
m[b'foo'] = want
self.assertEqual(want, m[b'foo'])
self.assertEqual([(b'bar/baz/qux.py', BIN_HASH_2),
(b'foo', BIN_HASH_1 + b'a')],
list(m.items()))
# Sometimes it even tries a 22-byte fake hash, but we can
# return 21 and it'll work out
m[b'foo'] = want + b'+'
self.assertEqual(want, m[b'foo'])
# make sure the suffix survives a copy
match = matchmod.match(b'', b'', [b're:foo'])
m2 = m.matches(match)
self.assertEqual(want, m2[b'foo'])
self.assertEqual(1, len(m2))
m2 = m.copy()
self.assertEqual(want, m2[b'foo'])
# suffix with iteration
self.assertEqual([(b'bar/baz/qux.py', BIN_HASH_2),
(b'foo', want)],
list(m.items()))
# shows up in diff
self.assertEqual({b'foo': ((want, f), (h, b''))}, m.diff(clean))
self.assertEqual({b'foo': ((h, b''), (want, f))}, clean.diff(m))
def testMatchException(self):
m = self.parsemanifest(A_SHORT_MANIFEST)
match = matchmod.match(b'', b'', [b're:.*'])
def filt(path):
if path == b'foo':
assert False
return True
match.matchfn = filt
with self.assertRaises(AssertionError):
m.matches(match)
def testRemoveItem(self):
m = self.parsemanifest(A_SHORT_MANIFEST)
del m[b'foo']
with self.assertRaises(KeyError):
m[b'foo']
self.assertEqual(1, len(m))
self.assertEqual(1, len(list(m)))
# now restore and make sure everything works right
m[b'foo'] = b'a' * 20
self.assertEqual(2, len(m))
self.assertEqual(2, len(list(m)))
def testManifestDiff(self):
MISSING = (None, b'')
addl = b'z-only-in-left\0' + HASH_1 + b'\n'
addr = b'z-only-in-right\0' + HASH_2 + b'x\n'
left = self.parsemanifest(
A_SHORT_MANIFEST.replace(HASH_1, HASH_3 + b'x') + addl)
right = self.parsemanifest(A_SHORT_MANIFEST + addr)
want = {
b'foo': ((BIN_HASH_3, b'x'),
(BIN_HASH_1, b'')),
b'z-only-in-left': ((BIN_HASH_1, b''), MISSING),
b'z-only-in-right': (MISSING, (BIN_HASH_2, b'x')),
}
self.assertEqual(want, left.diff(right))
want = {
b'bar/baz/qux.py': (MISSING, (BIN_HASH_2, b'l')),
b'foo': (MISSING, (BIN_HASH_3, b'x')),
b'z-only-in-left': (MISSING, (BIN_HASH_1, b'')),
}
self.assertEqual(want, self.parsemanifest(EMTPY_MANIFEST).diff(left))
want = {
b'bar/baz/qux.py': ((BIN_HASH_2, b'l'), MISSING),
b'foo': ((BIN_HASH_3, b'x'), MISSING),
b'z-only-in-left': ((BIN_HASH_1, b''), MISSING),
}
self.assertEqual(want, left.diff(self.parsemanifest(EMTPY_MANIFEST)))
copy = right.copy()
del copy[b'z-only-in-right']
del right[b'foo']
want = {
b'foo': (MISSING, (BIN_HASH_1, b'')),
b'z-only-in-right': ((BIN_HASH_2, b'x'), MISSING),
}
self.assertEqual(want, right.diff(copy))
short = self.parsemanifest(A_SHORT_MANIFEST)
pruned = short.copy()
del pruned[b'foo']
want = {
b'foo': ((BIN_HASH_1, b''), MISSING),
}
self.assertEqual(want, short.diff(pruned))
want = {
b'foo': (MISSING, (BIN_HASH_1, b'')),
}
self.assertEqual(want, pruned.diff(short))
want = {
b'bar/baz/qux.py': None,
b'foo': (MISSING, (BIN_HASH_1, b'')),
}
self.assertEqual(want, pruned.diff(short, clean=True))
def testReversedLines(self):
backwards = b''.join(
l + b'\n' for l in reversed(A_SHORT_MANIFEST.split(b'\n')) if l)
try:
self.parsemanifest(backwards)
self.fail('Should have raised ValueError')
except ValueError as v:
self.assertIn('Manifest lines not in sorted order.', str(v))
def testNoTerminalNewline(self):
try:
self.parsemanifest(A_SHORT_MANIFEST + b'wat')
self.fail('Should have raised ValueError')
except ValueError as v:
self.assertIn('Manifest did not end in a newline.', str(v))
def testNoNewLineAtAll(self):
try:
self.parsemanifest(b'wat')
self.fail('Should have raised ValueError')
except ValueError as v:
self.assertIn('Manifest did not end in a newline.', str(v))
def testHugeManifest(self):
m = self.parsemanifest(A_HUGE_MANIFEST)
self.assertEqual(HUGE_MANIFEST_ENTRIES, len(m))
self.assertEqual(len(m), len(list(m)))
def testMatchesMetadata(self):
'''Tests matches() for a few specific files to make sure that both
the set of files as well as their flags and nodeids are correct in
the resulting manifest.'''
m = self.parsemanifest(A_HUGE_MANIFEST)
match = matchmod.match(b'/', b'',
[b'file1', b'file200', b'file300'], exact=True)
m2 = m.matches(match)
w = (b'file1\0%sx\n'
b'file200\0%sl\n'
b'file300\0%s\n') % (HASH_2, HASH_1, HASH_1)
self.assertEqual(w, m2.text())
def testMatchesNonexistentFile(self):
'''Tests matches() for a small set of specific files, including one
nonexistent file to make sure in only matches against existing files.
'''
m = self.parsemanifest(A_DEEPER_MANIFEST)
match = matchmod.match(b'/', b'',
[b'a/b/c/bar.txt', b'a/b/d/qux.py',
b'readme.txt', b'nonexistent'],
exact=True)
m2 = m.matches(match)
self.assertEqual(
[b'a/b/c/bar.txt', b'a/b/d/qux.py', b'readme.txt'],
m2.keys())
def testMatchesNonexistentDirectory(self):
'''Tests matches() for a relpath match on a directory that doesn't
actually exist.'''
m = self.parsemanifest(A_DEEPER_MANIFEST)
match = matchmod.match(b'/', b'', [b'a/f'], default=b'relpath')
m2 = m.matches(match)
self.assertEqual([], m2.keys())
def testMatchesExactLarge(self):
'''Tests matches() for files matching a large list of exact files.
'''
m = self.parsemanifest(A_HUGE_MANIFEST)
flist = m.keys()[80:300]
match = matchmod.match(b'/', b'', flist, exact=True)
m2 = m.matches(match)
self.assertEqual(flist, m2.keys())
def testMatchesFull(self):
'''Tests matches() for what should be a full match.'''
m = self.parsemanifest(A_DEEPER_MANIFEST)
match = matchmod.match(b'/', b'', [b''])
m2 = m.matches(match)
self.assertEqual(m.keys(), m2.keys())
def testMatchesDirectory(self):
'''Tests matches() on a relpath match on a directory, which should
match against all files within said directory.'''
m = self.parsemanifest(A_DEEPER_MANIFEST)
match = matchmod.match(b'/', b'', [b'a/b'], default=b'relpath')
m2 = m.matches(match)
self.assertEqual([
b'a/b/c/bar.py', b'a/b/c/bar.txt', b'a/b/c/foo.py',
b'a/b/c/foo.txt',
b'a/b/d/baz.py', b'a/b/d/qux.py', b'a/b/d/ten.txt', b'a/b/dog.py',
b'a/b/fish.py'], m2.keys())
def testMatchesExactPath(self):
'''Tests matches() on an exact match on a directory, which should
result in an empty manifest because you can't perform an exact match
against a directory.'''
m = self.parsemanifest(A_DEEPER_MANIFEST)
match = matchmod.match(b'/', b'', [b'a/b'], exact=True)
m2 = m.matches(match)
self.assertEqual([], m2.keys())
def testMatchesCwd(self):
'''Tests matches() on a relpath match with the current directory ('.')
when not in the root directory.'''
m = self.parsemanifest(A_DEEPER_MANIFEST)
match = matchmod.match(b'/', b'a/b', [b'.'], default=b'relpath')
m2 = m.matches(match)
self.assertEqual([
b'a/b/c/bar.py', b'a/b/c/bar.txt', b'a/b/c/foo.py',
b'a/b/c/foo.txt', b'a/b/d/baz.py', b'a/b/d/qux.py',
b'a/b/d/ten.txt', b'a/b/dog.py', b'a/b/fish.py'], m2.keys())
def testMatchesWithPattern(self):
'''Tests matches() for files matching a pattern that reside
deeper than the specified directory.'''
m = self.parsemanifest(A_DEEPER_MANIFEST)
match = matchmod.match(b'/', b'', [b'a/b/*/*.txt'])
m2 = m.matches(match)
self.assertEqual(
[b'a/b/c/bar.txt', b'a/b/c/foo.txt', b'a/b/d/ten.txt'],
m2.keys())
class testmanifestdict(unittest.TestCase, basemanifesttests):
def parsemanifest(self, text):
return manifestmod.manifestdict(text)
def testObviouslyBogusManifest(self):
# This is a 163k manifest that came from oss-fuzz. It was a
# timeout there, but when run normally it doesn't seem to
# present any particular slowness.
data = zlib.decompress(
b'x\x9c\xed\xce;\n\x83\x00\x10\x04\xd0\x8deNa\x93~\xf1\x03\xc9q\xf4'
b'\x14\xeaU\xbdB\xda\xd4\xe6Cj\xc1FA\xde+\x86\xe9f\xa2\xfci\xbb\xfb'
b'\xa3\xef\xea\xba\xca\x7fk\x86q\x9a\xc6\xc8\xcc&\xb3\xcf\xf8\xb8|#'
b'\x8a9\x00\xd8\xe6v\xf4\x01N\xe1\n\x00\x00\x00\x00\x00\x00\x00\x00'
b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00'
b'\x00\x00\xc0\x8aey\x1d}\x01\xd8\xe0\xb9\xf3\xde\x1b\xcf\x17'
b'\xac\xbe')
with self.assertRaises(ValueError):
self.parsemanifest(data)
class testtreemanifest(unittest.TestCase, basemanifesttests):
def parsemanifest(self, text):
return manifestmod.treemanifest(b'', text)
def testWalkSubtrees(self):
m = self.parsemanifest(A_DEEPER_MANIFEST)
dirs = [s._dir for s in m.walksubtrees()]
self.assertEqual(
sorted([
b'', b'a/', b'a/c/', b'a/d/', b'a/b/', b'a/b/c/', b'a/b/d/']),
sorted(dirs)
)
match = matchmod.match(b'/', b'', [b'path:a/b/'])
dirs = [s._dir for s in m.walksubtrees(matcher=match)]
self.assertEqual(
sorted([b'a/b/', b'a/b/c/', b'a/b/d/']),
sorted(dirs)
)
if __name__ == '__main__':
silenttestrunner.main(__name__)