Commit 3696d569 authored by Jason Madden's avatar Jason Madden

Update changenotes now that sendall is faster. [skip ci]

parent db52e45c
...@@ -27,17 +27,13 @@ through 2.5.1, 2.6.0 and pre-release versions of 2.7.0. ...@@ -27,17 +27,13 @@ through 2.5.1, 2.6.0 and pre-release versions of 2.7.0.
to 2.7.0, signals could be delivered incorrectly or fail to be to 2.7.0, signals could be delivered incorrectly or fail to be
delivered during a blocking operation. delivered during a blocking operation.
- Overall performance seems to be quite acceptable with newer versions - Overall performance seems to be quite acceptable with newer versions
of PyPy. Things that are known or expected to be slower under PyPy of PyPy. The benchmarks distributed with gevent typically perform as
include the :mod:`c-ares resolver <gevent.resolver_ares>` and well or better on PyPy than on CPython. Things that are known or
:meth:`socket.socket.sendall`. In particular, expected to be (relatively) slower under PyPy include the
:meth:`socket.socket.sendall` can be `relatively slow`_ for large :mod:`c-ares resolver <gevent.resolver_ares>` and
transmissions. This can be mitigated by setting a larger write :class:`gevent.lock.Semaphore`. Whether or not these matter will
buffer on the socket, e.g, ``sock.setsockopt(socket.SOL_SOCKET,
socket.SO_SNDBUF, 1024*1024)`` Whether or not these matter will
depend on the workload of each application. depend on the workload of each application.
.. _relatively slow: https://bitbucket.org/pypy/pypy/issues/2091/non-blocking-socketsend-slow-gevent
Improved subprocess support Improved subprocess support
=========================== ===========================
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment