streamclone: use backgroundfilecloser (issue4889)
Closing files that have been appended to is slow on Windows/NTFS.
CloseHandle() calls on this platform often take 1-10ms - and that's
on my i7-6700K Skylake processor with a modern and fast SSD. Contrast
with other I/O operations, such as writing data, which take <100us.
This means that creating/appending thousands of files can add
significant overhead. For example, cloning mozilla-central creates
~232,000 revlog files. Assuming 1ms per CloseHandle(), that yields
232s (3:52) of wall time waiting for file closes!
The impact of this overhead can be measured most directly when applying
stream clone bundles. Applying these files is effectively uncompressing
a tar archive (read: it's very fast).
Using a RAM disk (read: no I/O wait), the difference in wall time for a
`hg debugapplystreamclonebundle` for a ~1731 MB mozilla-central bundle
between Windows and Linux from the same machine is drastic:
Linux: ~12.8s (128MB/s)
Windows: ~352.0s (4.7MB/s)
Windows is ~27.5x slower. Yikes!
After this patch:
Linux: ~12.8s (128MB/s)
Windows: ~102.1s (16.1MB/s)
Windows is now ~3.4x faster. Unfortunately, it is still ~8x slower than
Linux. Profiling reveals a few hot code paths that could likely be
improved. But those are for other patches.
This patch introduces test-clone-uncompressed.t because existing tests
of `clone --uncompressed` are scattered about and adding a variation for
background thread closing to e.g. test-http.t doesn't feel correct.
"""test line matching with some failing examples and some which warn
run-test.t only checks positive matches and can not see warnings
(both by design)
"""
from __future__ import print_function
import os, re
# this is hack to make sure no escape characters are inserted into the output
if 'TERM' in os.environ:
del os.environ['TERM']
import doctest
run_tests = __import__('run-tests')
def prn(ex):
m = ex.args[0]
if isinstance(m, str):
print(m)
else:
print(m.decode('utf-8'))
def lm(expected, output):
r"""check if output matches expected
does it generally work?
>>> lm(b'H*e (glob)\n', b'Here\n')
True
fail on bad test data
>>> try: lm(b'a\n',b'a')
... except AssertionError as ex: print(ex)
missing newline
>>> try: lm(b'single backslash\n', b'single \backslash\n')
... except AssertionError as ex: prn(ex)
single backslash or unknown char
"""
assert (expected.endswith(b'\n')
and output.endswith(b'\n')), 'missing newline'
assert not re.search(br'[^ \w\\/\r\n()*?]', expected + output), \
b'single backslash or unknown char'
match = run_tests.TTest.linematch(expected, output)
if isinstance(match, str):
return 'special: ' + match
elif isinstance(match, bytes):
return 'special: ' + match.decode('utf-8')
else:
return bool(match) # do not return match object
def wintests():
r"""test matching like running on windows
enable windows matching on any os
>>> _osaltsep = os.altsep
>>> os.altsep = True
valid match on windows
>>> lm(b'g/a*/d (glob)\n', b'g\\abc/d\n')
True
direct matching, glob unnecessary
>>> lm(b'g/b (glob)\n', b'g/b\n')
'special: -glob'
missing glob
>>> lm(b'/g/c/d/fg\n', b'\\g\\c\\d/fg\n')
'special: +glob'
restore os.altsep
>>> os.altsep = _osaltsep
"""
pass
def otherostests():
r"""test matching like running on non-windows os
disable windows matching on any os
>>> _osaltsep = os.altsep
>>> os.altsep = False
backslash does not match slash
>>> lm(b'h/a* (glob)\n', b'h\\ab\n')
False
direct matching glob can not be recognized
>>> lm(b'h/b (glob)\n', b'h/b\n')
True
missing glob can not not be recognized
>>> lm(b'/h/c/df/g/\n', b'\\h/c\\df/g\\\n')
False
restore os.altsep
>>> os.altsep = _osaltsep
"""
pass
if __name__ == '__main__':
doctest.testmod()