revlog: don't flush data file after every added revision
The current behavior of revlogs is to flush the data file when writing
data to it. Tracing system calls revealed that changegroup processing
incurred numerous write(2) calls for values much smaller than the
default buffer size (Python defaults to 4096, but it can be adjusted
based on detected block size at run time by CPython).
The reason we flush revlogs is so readers have all data available.
For example, the current code in revlog.py will re-open the revlog
file (instead of seeking an existing file handle) to read the text
of a revision. This happens when starting a new delta chain when
adding several revisions from changegroups, for example. Yes, this
is likely sub-optimal (we should probably be sharing file descriptors
between readers and writers to avoid the flushing and associated
overhead of re-opening files).
While flushing revlogs is necessary, it appears all callers are
diligent about flushing files before a read is performed (see
buildtext() in _addrevision()), making the flush in
_writeentry() redundant and unncessary. So, we remove it. In practice,
this means we incur a write(2) a) when the buffer is full (typically
4096 bytes) b) when a new delta chain is created rather than after
every added revision. This applies to every revlog, but by volume
it mostly impacts filelogs.
Removing the redundant flush from _writeentry() significantly
reduces the number of write(2) calls during changegroup processing on
my Linux machine. When applying a changegroup of the hg repo based on
my local repo, the total number of write(2) calls during application
of the mercurial/localrepo.py revlogs dropped from 1,320 to 217 with
this patch applied. Total I/O related system calls dropped from 1,577
to 474.
When unbundling a mozilla-central gzipped bundle (264,403 changesets
with 1,492,215 changes to 222,507 files), total write(2) calls
dropped from 1,252,881 to 827,106 and total system calls dropped from
3,601,259 to 3,178,636 - a reduction of 425,775!
While the system call reduction is significant, it appears
to have no impact on wall time on my Linux and Windows machines. Still,
fewer syscalls is fewer syscalls. Surely this can't hurt. If nothing
else, it makes examining remaining system call usage simpler and opens
the door to experimenting with the performance impact of different
buffer sizes.
# pager.py - display output using a pager
#
# Copyright 2008 David Soria Parra <dsp@php.net>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
#
# To load the extension, add it to your configuration file:
#
# [extension]
# pager =
#
# Run "hg help pager" to get info on configuration.
'''browse command output with an external pager
To set the pager that should be used, set the application variable::
[pager]
pager = less -FRX
If no pager is set, the pager extensions uses the environment variable
$PAGER. If neither pager.pager, nor $PAGER is set, no pager is used.
You can disable the pager for certain commands by adding them to the
pager.ignore list::
[pager]
ignore = version, help, update
You can also enable the pager only for certain commands using
pager.attend. Below is the default list of commands to be paged::
[pager]
attend = annotate, cat, diff, export, glog, log, qdiff
Setting pager.attend to an empty value will cause all commands to be
paged.
If pager.attend is present, pager.ignore will be ignored.
Lastly, you can enable and disable paging for individual commands with
the attend-<command> option. This setting takes precedence over
existing attend and ignore options and defaults::
[pager]
attend-cat = false
To ignore global commands like :hg:`version` or :hg:`help`, you have
to specify them in your user configuration file.
The --pager=... option can also be used to control when the pager is
used. Use a boolean value like yes, no, on, off, or use auto for
normal behavior.
'''
import atexit, sys, os, signal, subprocess
from mercurial import commands, dispatch, util, extensions, cmdutil
from mercurial.i18n import _
# Note for extension authors: ONLY specify testedwith = 'internal' for
# extensions which SHIP WITH MERCURIAL. Non-mainline extensions should
# be specifying the version(s) of Mercurial they are tested with, or
# leave the attribute unspecified.
testedwith = 'internal'
def _pagersubprocess(ui, p):
pager = subprocess.Popen(p, shell=True, bufsize=-1,
close_fds=util.closefds, stdin=subprocess.PIPE,
stdout=sys.stdout, stderr=sys.stderr)
stdout = os.dup(sys.stdout.fileno())
stderr = os.dup(sys.stderr.fileno())
os.dup2(pager.stdin.fileno(), sys.stdout.fileno())
if ui._isatty(sys.stderr):
os.dup2(pager.stdin.fileno(), sys.stderr.fileno())
@atexit.register
def killpager():
if util.safehasattr(signal, "SIGINT"):
signal.signal(signal.SIGINT, signal.SIG_IGN)
pager.stdin.close()
os.dup2(stdout, sys.stdout.fileno())
os.dup2(stderr, sys.stderr.fileno())
pager.wait()
def _runpager(ui, p):
_pagersubprocess(ui, p)
def uisetup(ui):
if '--debugger' in sys.argv or not ui.formatted():
return
def pagecmd(orig, ui, options, cmd, cmdfunc):
p = ui.config("pager", "pager", os.environ.get("PAGER"))
usepager = False
always = util.parsebool(options['pager'])
auto = options['pager'] == 'auto'
if not p:
pass
elif always:
usepager = True
elif not auto:
usepager = False
else:
attend = ui.configlist('pager', 'attend', attended)
ignore = ui.configlist('pager', 'ignore')
cmds, _ = cmdutil.findcmd(cmd, commands.table)
for cmd in cmds:
var = 'attend-%s' % cmd
if ui.config('pager', var):
usepager = ui.configbool('pager', var)
break
if (cmd in attend or
(cmd not in ignore and not attend)):
usepager = True
break
setattr(ui, 'pageractive', usepager)
if usepager:
ui.setconfig('ui', 'formatted', ui.formatted(), 'pager')
ui.setconfig('ui', 'interactive', False, 'pager')
if util.safehasattr(signal, "SIGPIPE"):
signal.signal(signal.SIGPIPE, signal.SIG_DFL)
_runpager(ui, p)
return orig(ui, options, cmd, cmdfunc)
# Wrap dispatch._runcommand after color is loaded so color can see
# ui.pageractive. Otherwise, if we loaded first, color's wrapped
# dispatch._runcommand would run without having access to ui.pageractive.
def afterloaded(loaded):
extensions.wrapfunction(dispatch, '_runcommand', pagecmd)
extensions.afterloaded('color', afterloaded)
def extsetup(ui):
commands.globalopts.append(
('', 'pager', 'auto',
_("when to paginate (boolean, always, auto, or never)"),
_('TYPE')))
attended = ['annotate', 'cat', 'diff', 'export', 'glog', 'log', 'qdiff']