py3: make stdout line-buffered if connected to a TTY
Status messages that are to be shown on the terminal should be written to the
file descriptor before anything further is done, to keep the user updated.
One common way to achieve this is to make stdout line-buffered if it is
connected to a TTY. This is done on Python 2 (except on Windows, where libc,
which the CPython 2 streams depend on, does not properly support this).
Python 3 rolls it own I/O streams. On Python 3, buffered binary streams can't be
set line-buffered. The previous code (added in 227ba1afcb65) incorrectly
assumed that on Python 3, pycompat.stdout (sys.stdout.buffer) is already
line-buffered. However the interpreter initializes it with a block-buffered
stream or an unbuffered stream (when the -u option or the PYTHONUNBUFFERED
environment variable is set), never with a line-buffered stream.
One example where the current behavior is unacceptable is when running
`hg pull https://www.mercurial-scm.org/repo/hg` on Python 3, where the line
"pulling from https://www.mercurial-scm.org/repo/hg" does not appear on the
terminal before the hg process blocks while waiting for the server.
Various approaches to fix this problem are possible, including:
1. Weaken the contract of procutil.stdout to not give any guarantees about
buffering behavior. In this case, users of procutil.stdout need to be
changed to do enough flushes. In particular,
1. either ui must insert enough flushes for ui.write() and friends, or
2. ui.write() and friends get split into flushing and fully buffered
methods, or
3. users of ui.write() and friends must flush explicitly.
2. Make stdout unbuffered.
3. Make stdout line-buffered. Since Python 3 does not natively support that for
binary streams, we must implement it ourselves.
(2.) is problematic because using unbuffered I/O changes the performance
characteristics significantly compared to line-buffered (which is used on
Python 2) and this would be a regression.
(1.2.) and (1.3) are a substantial amount of work. It’s unclear whether the
added complexity would be justified, given that raw performance doesn’t matter
that much when writing to a terminal much faster than the user could read it.
(1.1.) pushes complexity into the ui class instead of separating the concern of
how stdout is buffered. Other users of procutil.stdout would still need to take
care of the flushes.
This patch implements (3.). The general performance considerations are very
similar to (1.1.). The extra method invocation and method forwarding add a
little more overhead if the class is used. In exchange, it doesn’t add overhead
if not used.
For the benchmarks, I compared the previous implementation (incorrect on Python
3), (1.1.), (3.) and (2.). The command was chosen so that the streams were
configured as if they were writing to a TTY, but actually write to a pager,
which is also the default:
HGRCPATH=/dev/null python3 ./hg --cwd ~/vcs/mozilla-central --time --pager yes --config pager.pager='cat > /dev/null' status --all
previous:
time: real 7.880 secs (user 7.290+0.050 sys 0.580+0.170)
time: real 7.830 secs (user 7.220+0.070 sys 0.590+0.140)
time: real 7.800 secs (user 7.210+0.050 sys 0.570+0.170)
(1.1.) using Yuya Nishihara’s patch:
time: real 9.860 secs (user 8.670+0.350 sys 1.160+0.830)
time: real 9.540 secs (user 8.430+0.370 sys 1.100+0.770)
time: real 9.830 secs (user 8.630+0.370 sys 1.180+0.840)
(3.) using this patch:
time: real 9.580 secs (user 8.480+0.350 sys 1.090+0.770)
time: real 9.670 secs (user 8.480+0.330 sys 1.170+0.860)
time: real 9.640 secs (user 8.500+0.350 sys 1.130+0.810)
(2.) using a previous patch by me:
time: real 10.480 secs (user 8.850+0.720 sys 1.590+1.500)
time: real 10.490 secs (user 8.750+0.750 sys 1.710+1.470)
time: real 10.240 secs (user 8.600+0.700 sys 1.590+1.510)
As expected, there’s no difference on Python 2, as exactly the same code paths
are used:
previous:
time: real 6.950 secs (user 5.870+0.330 sys 1.070+0.770)
time: real 7.040 secs (user 6.040+0.360 sys 0.980+0.750)
time: real 7.070 secs (user 5.950+0.360 sys 1.100+0.760)
this patch:
time: real 7.010 secs (user 5.900+0.390 sys 1.070+0.730)
time: real 7.000 secs (user 5.850+0.350 sys 1.120+0.760)
time: real 7.000 secs (user 5.790+0.380 sys 1.170+0.710)
Create user cache directory
$ USERCACHE=`pwd`/cache; export USERCACHE
$ cat <<EOF >> ${HGRCPATH}
> [extensions]
> hgext.largefiles=
> [largefiles]
> usercache=${USERCACHE}
> EOF
$ mkdir -p ${USERCACHE}
Create source repo, and commit adding largefile.
$ hg init src
$ cd src
$ echo large > large
$ hg add --large large
$ hg commit -m 'add largefile'
$ hg rm large
$ hg commit -m 'branchhead without largefile' large
$ hg up -qr 0
$ rm large
$ echo "0000000000000000000000000000000000000000" > .hglf/large
$ hg commit -m 'commit missing file with corrupt standin' large
abort: large: file not found!
[255]
$ hg up -Cqr 0
$ cd ..
Discard all cached largefiles in USERCACHE
$ rm -rf ${USERCACHE}
Create mirror repo, and pull from source without largefile:
"pull" is used instead of "clone" for suppression of (1) updating to
tip (= caching largefile from source repo), and (2) recording source
repo as "default" path in .hg/hgrc.
$ hg init mirror
$ cd mirror
$ hg pull ../src
pulling from ../src
requesting all changes
adding changesets
adding manifests
adding file changes
added 2 changesets with 1 changes to 1 files
new changesets eb85d9124f3f:26c18ce05e4e
(run 'hg update' to get a working copy)
Update working directory to "tip", which requires largefile("large"),
but there is no cache file for it. So, hg must treat it as
"missing"(!) file.
$ hg update -r0
getting changed largefiles
large: largefile 7f7097b041ccf68cc5561e9600da4655d21c6d18 not available from file:/*/$TESTTMP/mirror (glob)
0 largefiles updated, 0 removed
1 files updated, 0 files merged, 0 files removed, 0 files unresolved
$ hg status
! large
Update working directory to null: this cleanup .hg/largefiles/dirstate
$ hg update null
getting changed largefiles
0 largefiles updated, 0 removed
0 files updated, 0 files merged, 1 files removed, 0 files unresolved
Update working directory to tip, again.
$ hg update -r0
getting changed largefiles
large: largefile 7f7097b041ccf68cc5561e9600da4655d21c6d18 not available from file:/*/$TESTTMP/mirror (glob)
0 largefiles updated, 0 removed
1 files updated, 0 files merged, 0 files removed, 0 files unresolved
$ hg status
! large
$ cd ..
Verify that largefiles from pulled branchheads are fetched, also to an empty repo
$ hg init mirror2
$ hg -R mirror2 pull src -r0
pulling from src
adding changesets
adding manifests
adding file changes
added 1 changesets with 1 changes to 1 files
new changesets eb85d9124f3f
(run 'hg update' to get a working copy)
#if unix-permissions
Portable way to print file permissions:
$ cat > ls-l.py <<EOF
> #!$PYTHON
> from __future__ import absolute_import, print_function
> import os
> import sys
> path = sys.argv[1]
> print('%03o' % (os.lstat(path).st_mode & 0o777))
> EOF
$ chmod +x ls-l.py
Test that files in .hg/largefiles inherit mode from .hg/store, not
from file in working copy:
$ cd src
$ chmod 750 .hg/store
$ chmod 660 large
$ echo change >> large
$ hg commit -m change
created new head
$ ../ls-l.py .hg/largefiles/e151b474069de4ca6898f67ce2f2a7263adf8fea
640
Test permission of with files in .hg/largefiles created by update:
$ cd ../mirror
$ rm -r "$USERCACHE" .hg/largefiles # avoid links
$ chmod 750 .hg/store
$ hg pull ../src --update -q
$ ../ls-l.py .hg/largefiles/e151b474069de4ca6898f67ce2f2a7263adf8fea
640
Test permission of files created by push:
$ hg serve -R ../src -d -p $HGPORT --pid-file hg.pid \
> --config "web.allow_push=*" --config web.push_ssl=no
$ cat hg.pid >> $DAEMON_PIDS
$ echo change >> large
$ hg commit -m change
$ rm -r "$USERCACHE"
$ hg push -q http://localhost:$HGPORT/
$ ../ls-l.py ../src/.hg/largefiles/b734e14a0971e370408ab9bce8d56d8485e368a9
640
$ cd ..
#endif
Test issue 4053 (remove --after on a deleted, uncommitted file shouldn't say
it is missing, but a remove on a nonexistent unknown file still should. Same
for a forget.)
$ cd src
$ touch x
$ hg add x
$ mv x y
$ hg remove -A x y ENOENT
ENOENT: * (glob)
not removing y: file is untracked
[1]
$ hg add y
$ mv y z
$ hg forget y z ENOENT
ENOENT: * (glob)
not removing z: file is already untracked
[1]
Largefiles are accessible from the share's store
$ cd ..
$ hg share -q src share_dst --config extensions.share=
$ hg -R share_dst update -r0
getting changed largefiles
1 largefiles updated, 0 removed
1 files updated, 0 files merged, 0 files removed, 0 files unresolved
$ echo modified > share_dst/large
$ hg -R share_dst ci -m modified
created new head
Only dirstate is in the local store for the share, and the largefile is in the
share source's local store. Avoid the extra largefiles added in the unix
conditional above.
$ hash=`hg -R share_dst cat share_dst/.hglf/large`
$ echo $hash
e2fb5f2139d086ded2cb600d5a91a196e76bf020
$ find share_dst/.hg/largefiles/* | sort
share_dst/.hg/largefiles/dirstate
$ find src/.hg/largefiles/* | egrep "(dirstate|$hash)" | sort
src/.hg/largefiles/dirstate
src/.hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020
Verify that backwards compatibility is maintained for old storage layout
$ mv src/.hg/largefiles/$hash share_dst/.hg/largefiles
$ hg verify --quiet --lfa -R share_dst --config largefiles.usercache=
Inject corruption into the largefiles store and see how update handles that:
$ cd src
$ hg up -qC tip
$ cat large
modified
$ rm large
$ cat .hglf/large
e2fb5f2139d086ded2cb600d5a91a196e76bf020
$ mv .hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020 ..
$ echo corruption > .hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020
$ hg up -C
getting changed largefiles
large: data corruption in $TESTTMP/src/.hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020 with hash 6a7bb2556144babe3899b25e5428123735bb1e27
0 largefiles updated, 0 removed
0 files updated, 0 files merged, 0 files removed, 0 files unresolved
updated to "cd24c147f45c: modified"
[12] other heads for branch "default" (re)
$ hg st
! large
? z
$ rm .hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020
#if serve
Test coverage of error handling from putlfile:
$ mkdir $TESTTMP/mirrorcache
$ hg serve -R ../mirror -d -p $HGPORT1 --pid-file hg.pid --config largefiles.usercache=$TESTTMP/mirrorcache
$ cat hg.pid >> $DAEMON_PIDS
$ hg push http://localhost:$HGPORT1 -f --config files.usercache=nocache
pushing to http://localhost:$HGPORT1/
searching for changes
abort: remotestore: could not open file $TESTTMP/src/.hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020: HTTP Error 403: ssl required
[255]
$ rm .hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020
Test coverage of 'missing from store':
$ hg serve -R ../mirror -d -p $HGPORT2 --pid-file hg.pid --config largefiles.usercache=$TESTTMP/mirrorcache --config "web.allow_push=*" --config web.push_ssl=no
$ cat hg.pid >> $DAEMON_PIDS
$ hg push http://localhost:$HGPORT2 -f --config largefiles.usercache=nocache
pushing to http://localhost:$HGPORT2/
searching for changes
abort: largefile e2fb5f2139d086ded2cb600d5a91a196e76bf020 missing from store (needs to be uploaded)
[255]
Verify that --lfrev controls which revisions are checked for largefiles to push
$ hg push http://localhost:$HGPORT2 -f --config largefiles.usercache=nocache --lfrev tip
pushing to http://localhost:$HGPORT2/
searching for changes
abort: largefile e2fb5f2139d086ded2cb600d5a91a196e76bf020 missing from store (needs to be uploaded)
[255]
$ hg push http://localhost:$HGPORT2 -f --config largefiles.usercache=nocache --lfrev null
pushing to http://localhost:$HGPORT2/
searching for changes
remote: adding changesets
remote: adding manifests
remote: adding file changes
remote: added 1 changesets with 1 changes to 1 files (+1 heads)
#endif