view tests/test-convert-darcs.t @ 35793:4fb2bb61597c

bundle2: increase payload part chunk size to 32kb

Bundle2 payload parts are framed chunks. Esentially, we obtain
data in equal size chunks of size `preferedchunksize` and emit those
to a generator. That generator is fed into a compressor (which can
be the no-op compressor, which just re-emits the generator). And
the output from the compressor likely goes to a file descriptor
or socket.

What this means is that small chunk sizes create more Python objects
and Python function calls than larger chunk sizes. And as we know,
Python object and function call overhead in performance sensitive
code matters (at least with CPython).

This commit increases the bundle2 part payload chunk size from 4k
to 32k. Practically speaking, this means that the chunks we feed
into a compressor (implemented in C code) or feed directly into a
file handle or socket write() are larger. It's possible the chunks
might be larger than what the receiver can handle in one logical
operation. But at that point, we're in C code, which is much more
efficient at dealing with splitting up the chunk and making multiple
function calls than Python is.

A downside to larger chunks is that the receiver has to wait for that
much data to arrive (either raw or from a decompressor) before it
can process the chunk. But 32kb still feels like a small buffer to
have to wait for. And in many cases, the client will convert from
8 read(4096) to 1 read(32768). That's happening in Python land. So
we cut down on the number of Python objects and function calls,
making the client faster as well. I don't think there are any
significant concerns to increasing the payload chunk size to 32kb.

The impact of this change on performance significant. Using `curl`
to obtain a stream clone bundle2 payload from a server on localhost
serving the mozilla-unified repository:

before: 20.78 user; 7.71 system;  80.5 MB/s
after:  13.90 user; 3.51 system; 132 MB/s
legacy:  9.72 user; 8.16 system; 132 MB/s

bundle2 stream clone generation is still more resource intensive than
legacy stream clone (that's likely because of the use of a
util.chunkbuffer). But the throughput is the same. We might
be in territory we're this is effectively a benchmark of the
networking stack or Python's syscall throughput.

From the client perspective, `hg clone -U --stream`:

before: 33.50 user; 7.95 system; 53.3 MB/s
after:  22.82 user; 7.33 system; 72.7 MB/s
legacy: 29.96 user; 7.94 system; 58.0 MB/s

And for `hg clone --stream` with a working directory update of
~230k files:

after:  119.55 user; 26.47 system; 0:57.08 wall
legacy: 126.98 user; 26.94 system; 1:05.56 wall

So, it appears that bundle2's stream clone is now definitively faster
than legacy stream clone!

Differential Revision: https://phab.mercurial-scm.org/D1932

#require darcs

  $ echo "[extensions]" >> $HGRCPATH
  $ echo "convert=" >> $HGRCPATH
  $ DARCS_EMAIL='[e-mail]'; export DARCS_EMAIL

initialize darcs repo

  $ mkdir darcs-repo
  $ cd darcs-repo
  $ darcs init -q
  $ echo a > a
  $ darcs record -a -l -m p0
  Finished recording patch 'p0'
  $ cd ..

branch and update

  $ darcs get -q darcs-repo darcs-clone >/dev/null
  $ cd darcs-clone
  $ echo c >> a
  $ echo c > c
  $ darcs record -a -l -m p1.1
  Finished recording patch 'p1.1'
  $ cd ..

skip if we can't import elementtree

  $ if hg convert darcs-repo darcs-dummy 2>&1 | grep ElementTree > /dev/null; then
  >     echo 'skipped: missing feature: elementtree module'
  >     exit 80
  > fi

update source

  $ cd darcs-repo
  $ echo b >> a
  $ echo b > b
  $ darcs record -a -l -m p1.2
  Finished recording patch 'p1.2'

  $ darcs pull -q -a --no-set-default ../darcs-clone
  Backing up ./a(*) (glob)
  We have conflicts in the following files:
  ./a
   (?)
  $ sleep 1
  $ echo e > a
  $ echo f > f
  $ mkdir dir
  $ echo d > dir/d
  $ echo d > dir/d2
  $ darcs record -a -l -m p2
  Finished recording patch 'p2'

test file and directory move

  $ darcs mv -q f ff

Test remove + move

  $ darcs remove -q dir/d2
  $ rm dir/d2
  $ darcs mv -q dir dir2
  $ darcs record -a -l -m p3
  Finished recording patch 'p3'

The converter does not currently handle patch conflicts very well.
When they occur, it reverts *all* changes and moves forward,
letting the conflict resolving patch fix collisions.
Unfortunately, non-conflicting changes, like the addition of the
"c" file in p1.1 patch are reverted too.
Just to say that manifest not listing "c" here is a bug.

  $ cd ..
  $ hg convert darcs-repo darcs-repo-hg
  initializing destination darcs-repo-hg repository
  scanning source...
  sorting...
  converting...
  4 p0
  3 p1.2
  2 p1.1
  1 p2
  0 p3
  $ hg log -R darcs-repo-hg -g --template '{rev} "{desc|firstline}" ({author}) files: {files}\n' "$@"
  4 "p3" ([e-mail]) files: dir/d dir/d2 dir2/d f ff
  3 "p2" ([e-mail]) files: a dir/d dir/d2 f
  2 "p1.1" ([e-mail]) files: 
  1 "p1.2" ([e-mail]) files: a b
  0 "p0" ([e-mail]) files: a

  $ hg up -q -R darcs-repo-hg
  $ hg -R darcs-repo-hg manifest --debug
  7225b30cdf38257d5cc7780772c051b6f33e6d6b 644   a
  1e88685f5ddec574a34c70af492f95b6debc8741 644   b
  37406831adc447ec2385014019599dfec953c806 644   dir2/d
  b783a337463792a5c7d548ad85a7d3253c16ba8c 644   ff

#if no-outer-repo

try converting darcs1 repository

  $ hg clone -q "$TESTDIR/bundles/darcs1.hg" darcs
  $ hg convert -s darcs darcs/darcs1 2>&1 | grep darcs-1.0
  darcs-1.0 repository format is unsupported, please upgrade

#endif