##// END OF EJS Templates
phases: large rewrite on retract boundary...
phases: large rewrite on retract boundary The new code is still pure Python, so we still have room to going significantly faster. However its complexity of the complex part is `O(|[min_new_draft, tip]|)` instead of `O(|[min_draft, tip]|` which should help tremendously one repository with old draft (like mercurial-devel or mozilla-try). This is especially useful as the most common "retract boundary" operation happens when we commit/rewrite new drafts or when we push new draft to a non-publishing server. In this case, the smallest new_revs is very close to the tip and there is very few work to do. A few smaller optimisation could be done for these cases and will be introduced in later changesets. We still have iterate over large sets of roots, but this is already a great improvement for a very small amount of work. We gather information on the affected changeset as we go as we can put it to use in the next changesets. This extra data collection might slowdown the `register_new` case a bit, however for register_new, it should not really matters. The set of new nodes is either small, so the impact is negligible, or the set of new nodes is large, and the amount of work to do to had them will dominate the overhead the collecting information in `changed_revs`. As this new code compute the changes on the fly, it unlock other interesting improvement to be done in later changeset.

File last commit:

r49730:6000f5b2 default
r52302:2f39c7ae default
Show More
timestamp.py
127 lines | 3.8 KiB | text/x-python | PythonLexer
Simon Sapin
dirstate: store mtimes with nanosecond precision in memory...
r49079 # Copyright Mercurial Contributors
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
Simon Sapin
dirstate: ignore sub-second component when either is zero in mtime...
r49081 import functools
dirstate: move "get fs now" in the timestamp utility module...
r49202 import os
Simon Sapin
dirstate: store mtimes with nanosecond precision in memory...
r49079 import stat
dirstate: drop comparison primitive on the timestamp class...
r49226 from .. import error
Simon Sapin
dirstate: store mtimes with nanosecond precision in memory...
r49079
rangemask = 0x7FFFFFFF
Simon Sapin
dirstate: ignore sub-second component when either is zero in mtime...
r49081 @functools.total_ordering
Simon Sapin
dirstate: store mtimes with nanosecond precision in memory...
r49079 class timestamp(tuple):
"""
Simon Sapin
dirstate: ignore sub-second component when either is zero in mtime...
r49081 A Unix timestamp with optional nanoseconds precision,
Simon Sapin
dirstate: store mtimes with nanosecond precision in memory...
r49079 modulo 2**31 seconds.
Simon Sapin
dirstate: Document Timestamp.second_ambiguous...
r49271 A 3-tuple containing:
Simon Sapin
dirstate: store mtimes with nanosecond precision in memory...
r49079
`truncated_seconds`: seconds since the Unix epoch,
truncated to its lower 31 bits
`subsecond_nanoseconds`: number of nanoseconds since `truncated_seconds`.
Simon Sapin
dirstate: ignore sub-second component when either is zero in mtime...
r49081 When this is zero, the sub-second precision is considered unknown.
Simon Sapin
dirstate: Document Timestamp.second_ambiguous...
r49271
`second_ambiguous`: whether this timestamp is still "reliable"
(see `reliable_mtime_of`) if we drop its sub-second component.
Simon Sapin
dirstate: store mtimes with nanosecond precision in memory...
r49079 """
def __new__(cls, value):
dirstate-item: add a "second_ambiguous` flag in the mtime tuple...
r49227 truncated_seconds, subsec_nanos, second_ambiguous = value
value = (truncated_seconds & rangemask, subsec_nanos, second_ambiguous)
Simon Sapin
dirstate: store mtimes with nanosecond precision in memory...
r49079 return super(timestamp, cls).__new__(cls, value)
Simon Sapin
dirstate: ignore sub-second component when either is zero in mtime...
r49081 def __eq__(self, other):
dirstate: drop comparison primitive on the timestamp class...
r49226 raise error.ProgrammingError(
'timestamp should never be compared directly'
Simon Sapin
dirstate: ignore sub-second component when either is zero in mtime...
r49081 )
def __gt__(self, other):
dirstate: drop comparison primitive on the timestamp class...
r49226 raise error.ProgrammingError(
'timestamp should never be compared directly'
)
Simon Sapin
dirstate: ignore sub-second component when either is zero in mtime...
r49081
Simon Sapin
dirstate: store mtimes with nanosecond precision in memory...
r49079
dirstate: move "get fs now" in the timestamp utility module...
r49202 def get_fs_now(vfs):
"""return a timestamp for "now" in the current vfs
This will raise an exception if no temporary files could be created.
"""
tmpfd, tmpname = vfs.mkstemp()
try:
return mtime_of(os.fstat(tmpfd))
finally:
os.close(tmpfd)
vfs.unlink(tmpname)
Simon Sapin
dirstate: store mtimes with nanosecond precision in memory...
r49079 def zero():
"""
Returns the `timestamp` at the Unix epoch.
"""
return tuple.__new__(timestamp, (0, 0))
def mtime_of(stat_result):
"""
Takes an `os.stat_result`-like object and returns a `timestamp` object
for its modification time.
"""
Simon Sapin
dirstate-v2: actually use sub-second mtime precision...
r49082 try:
# TODO: add this attribute to `osutil.stat` objects,
# see `mercurial/cext/osutil.c`.
#
# This attribute is also not available on Python 2.
nanos = stat_result.st_mtime_ns
except AttributeError:
# https://docs.python.org/2/library/os.html#os.stat_float_times
# "For compatibility with older Python versions,
# accessing stat_result as a tuple always returns integers."
secs = stat_result[stat.ST_MTIME]
Simon Sapin
dirstate: store mtimes with nanosecond precision in memory...
r49079
Simon Sapin
dirstate-v2: actually use sub-second mtime precision...
r49082 subsec_nanos = 0
else:
billion = int(1e9)
secs = nanos // billion
subsec_nanos = nanos % billion
Simon Sapin
dirstate: store mtimes with nanosecond precision in memory...
r49079
dirstate-item: add a "second_ambiguous` flag in the mtime tuple...
r49227 return timestamp((secs, subsec_nanos, False))
status: move the boundary comparison logic within the timestamp module...
r49224
def reliable_mtime_of(stat_result, present_mtime):
Simon Sapin
dirstate: Document Timestamp.second_ambiguous...
r49271 """Same as `mtime_of`, but return `None` or a `Timestamp` with
`second_ambiguous` set if the date might be ambiguous.
status: move the boundary comparison logic within the timestamp module...
r49224
A modification time is reliable if it is older than "present_time" (or
Simon Sapin
rhg: Update the dirstate on disk after status...
r49250 sufficiently in the future).
status: move the boundary comparison logic within the timestamp module...
r49224
Otherwise a concurrent modification might happens with the same mtime.
"""
file_mtime = mtime_of(stat_result)
file_second = file_mtime[0]
status: keep second-ambiguous mtimes during fixup...
r49232 file_ns = file_mtime[1]
status: move the boundary comparison logic within the timestamp module...
r49224 boundary_second = present_mtime[0]
status: keep second-ambiguous mtimes during fixup...
r49232 boundary_ns = present_mtime[1]
status: move the boundary comparison logic within the timestamp module...
r49224 # If the mtime of the ambiguous file is younger (or equal) to the starting
# point of the `status` walk, we cannot garantee that another, racy, write
# will not happen right after with the same mtime and we cannot cache the
# information.
#
status: keep second-ambiguous mtimes during fixup...
r49232 # However if the mtime is far away in the future, this is likely some
status: move the boundary comparison logic within the timestamp module...
r49224 # mismatch between the current clock and previous file system operation. So
# mtime more than one days in the future are considered fine.
status: keep second-ambiguous mtimes during fixup...
r49232 if boundary_second == file_second:
if file_ns and boundary_ns:
if file_ns < boundary_ns:
return timestamp((file_second, file_ns, True))
return None
elif boundary_second < file_second < (3600 * 24 + boundary_second):
status: move the boundary comparison logic within the timestamp module...
r49224 return None
else:
return file_mtime