view tests/test-cbor.py @ 38732:be4984261611

merge: mark file gets as not thread safe (issue5933)

In default installs, this has the effect of disabling the thread-based
worker on Windows when manifesting files in the working directory. My
measurements have shown that with revlog-based repositories, Mercurial
spends a lot of CPU time in revlog code resolving file data. This ends
up incurring a lot of context switching across threads and slows down
`hg update` operations when going from an empty working directory to
the tip of the repo.

On mozilla-unified (246,351 files) on an i7-6700K (4+4 CPUs):

before: 487s wall
after:  360s wall (equivalent to worker.enabled=false)
cpus=2: 379s wall

Even with only 2 threads, the thread pool is still slower.

The introduction of the thread-based worker (02b36e860e0b) states that
it resulted in a "~50%" speedup for `hg sparse --enable-profile` and
`hg sparse --disable-profile`. This disagrees with my measurement
above. I theorize a few reasons for this:

1) Removal of files from the working directory is I/O - not CPU - bound
   and should benefit from a thread pool (unless I/O is insanely fast
   and the GIL release is near instantaneous). So tests like `hg sparse
   --enable-profile` may exercise deletion throughput and aren't good
   benchmarks for worker tasks that are CPU heavy.
2) The patch was authored by someone at Facebook. The results were
   likely measured against a repository using remotefilelog. And I
   believe that revision retrieval during working directory updates with
   remotefilelog will often use a remote store, thus being I/O and not
   CPU bound. This probably resulted in an overstated performance gain.

Since there appears to be a need to enable the thread-based worker with
some stores, I've made the flagging of file gets as thread safe
configurable. I've made it experimental because I don't want to formalize
a boolean flag for this option and because this attribute is best
captured against the store implementation. But we don't have a proper
store API for this yet. I'd rather cross this bridge later.

It is possible there are revlog-based repositories that do benefit from
a thread-based worker. I didn't do very comprehensive testing. If there
are, we may want to devise a more proper algorithm for whether to use
the thread-based worker, including possibly config options to limit the
number of threads to use. But until I see evidence that justifies
complexity, simplicity wins.

Differential Revision: https://phab.mercurial-scm.org/D3963

from __future__ import absolute_import

import io
import unittest

from mercurial.thirdparty import (
    cbor,
)
from mercurial.utils import (
    cborutil,
)

def loadit(it):
    return cbor.loads(b''.join(it))

class BytestringTests(unittest.TestCase):
    def testsimple(self):
        self.assertEqual(
            list(cborutil.streamencode(b'foobar')),
            [b'\x46', b'foobar'])

        self.assertEqual(
            loadit(cborutil.streamencode(b'foobar')),
            b'foobar')

    def testlong(self):
        source = b'x' * 1048576

        self.assertEqual(loadit(cborutil.streamencode(source)), source)

    def testfromiter(self):
        # This is the example from RFC 7049 Section 2.2.2.
        source = [b'\xaa\xbb\xcc\xdd', b'\xee\xff\x99']

        self.assertEqual(
            list(cborutil.streamencodebytestringfromiter(source)),
            [
                b'\x5f',
                b'\x44',
                b'\xaa\xbb\xcc\xdd',
                b'\x43',
                b'\xee\xff\x99',
                b'\xff',
            ])

        self.assertEqual(
            loadit(cborutil.streamencodebytestringfromiter(source)),
            b''.join(source))

    def testfromiterlarge(self):
        source = [b'a' * 16, b'b' * 128, b'c' * 1024, b'd' * 1048576]

        self.assertEqual(
            loadit(cborutil.streamencodebytestringfromiter(source)),
            b''.join(source))

    def testindefinite(self):
        source = b'\x00\x01\x02\x03' + b'\xff' * 16384

        it = cborutil.streamencodeindefinitebytestring(source, chunksize=2)

        self.assertEqual(next(it), b'\x5f')
        self.assertEqual(next(it), b'\x42')
        self.assertEqual(next(it), b'\x00\x01')
        self.assertEqual(next(it), b'\x42')
        self.assertEqual(next(it), b'\x02\x03')
        self.assertEqual(next(it), b'\x42')
        self.assertEqual(next(it), b'\xff\xff')

        dest = b''.join(cborutil.streamencodeindefinitebytestring(
            source, chunksize=42))
        self.assertEqual(cbor.loads(dest), source)

    def testreadtoiter(self):
        source = io.BytesIO(b'\x5f\x44\xaa\xbb\xcc\xdd\x43\xee\xff\x99\xff')

        it = cborutil.readindefinitebytestringtoiter(source)
        self.assertEqual(next(it), b'\xaa\xbb\xcc\xdd')
        self.assertEqual(next(it), b'\xee\xff\x99')

        with self.assertRaises(StopIteration):
            next(it)

class IntTests(unittest.TestCase):
    def testsmall(self):
        self.assertEqual(list(cborutil.streamencode(0)), [b'\x00'])
        self.assertEqual(list(cborutil.streamencode(1)), [b'\x01'])
        self.assertEqual(list(cborutil.streamencode(2)), [b'\x02'])
        self.assertEqual(list(cborutil.streamencode(3)), [b'\x03'])
        self.assertEqual(list(cborutil.streamencode(4)), [b'\x04'])

    def testnegativesmall(self):
        self.assertEqual(list(cborutil.streamencode(-1)), [b'\x20'])
        self.assertEqual(list(cborutil.streamencode(-2)), [b'\x21'])
        self.assertEqual(list(cborutil.streamencode(-3)), [b'\x22'])
        self.assertEqual(list(cborutil.streamencode(-4)), [b'\x23'])
        self.assertEqual(list(cborutil.streamencode(-5)), [b'\x24'])

    def testrange(self):
        for i in range(-70000, 70000, 10):
            self.assertEqual(
                b''.join(cborutil.streamencode(i)),
                cbor.dumps(i))

class ArrayTests(unittest.TestCase):
    def testempty(self):
        self.assertEqual(list(cborutil.streamencode([])), [b'\x80'])
        self.assertEqual(loadit(cborutil.streamencode([])), [])

    def testbasic(self):
        source = [b'foo', b'bar', 1, -10]

        self.assertEqual(list(cborutil.streamencode(source)), [
            b'\x84', b'\x43', b'foo', b'\x43', b'bar', b'\x01', b'\x29'])

    def testemptyfromiter(self):
        self.assertEqual(b''.join(cborutil.streamencodearrayfromiter([])),
                         b'\x9f\xff')

    def testfromiter1(self):
        source = [b'foo']

        self.assertEqual(list(cborutil.streamencodearrayfromiter(source)), [
            b'\x9f',
            b'\x43', b'foo',
            b'\xff',
        ])

        dest = b''.join(cborutil.streamencodearrayfromiter(source))
        self.assertEqual(cbor.loads(dest), source)

    def testtuple(self):
        source = (b'foo', None, 42)

        self.assertEqual(cbor.loads(b''.join(cborutil.streamencode(source))),
                         list(source))

class SetTests(unittest.TestCase):
    def testempty(self):
        self.assertEqual(list(cborutil.streamencode(set())), [
            b'\xd9\x01\x02',
            b'\x80',
        ])

    def testset(self):
        source = {b'foo', None, 42}

        self.assertEqual(cbor.loads(b''.join(cborutil.streamencode(source))),
                         source)

class BoolTests(unittest.TestCase):
    def testbasic(self):
        self.assertEqual(list(cborutil.streamencode(True)),  [b'\xf5'])
        self.assertEqual(list(cborutil.streamencode(False)), [b'\xf4'])

        self.assertIs(loadit(cborutil.streamencode(True)), True)
        self.assertIs(loadit(cborutil.streamencode(False)), False)

class NoneTests(unittest.TestCase):
    def testbasic(self):
        self.assertEqual(list(cborutil.streamencode(None)), [b'\xf6'])

        self.assertIs(loadit(cborutil.streamencode(None)), None)

class MapTests(unittest.TestCase):
    def testempty(self):
        self.assertEqual(list(cborutil.streamencode({})), [b'\xa0'])
        self.assertEqual(loadit(cborutil.streamencode({})), {})

    def testemptyindefinite(self):
        self.assertEqual(list(cborutil.streamencodemapfromiter([])), [
            b'\xbf', b'\xff'])

        self.assertEqual(loadit(cborutil.streamencodemapfromiter([])), {})

    def testone(self):
        source = {b'foo': b'bar'}
        self.assertEqual(list(cborutil.streamencode(source)), [
            b'\xa1', b'\x43', b'foo', b'\x43', b'bar'])

        self.assertEqual(loadit(cborutil.streamencode(source)), source)

    def testmultiple(self):
        source = {
            b'foo': b'bar',
            b'baz': b'value1',
        }

        self.assertEqual(loadit(cborutil.streamencode(source)), source)

        self.assertEqual(
            loadit(cborutil.streamencodemapfromiter(source.items())),
            source)

    def testcomplex(self):
        source = {
            b'key': 1,
            2: -10,
        }

        self.assertEqual(loadit(cborutil.streamencode(source)),
                         source)

        self.assertEqual(
            loadit(cborutil.streamencodemapfromiter(source.items())),
            source)

if __name__ == '__main__':
    import silenttestrunner
    silenttestrunner.main(__name__)