From: Joseph Jones <joeyi5216@...>
Date: 2015-12-17T21:06:57-07:00
Subject: [ruby-core:72291] Re: [Ruby trunk - Feature #11339] [Open] [PATCH] io.c: avoid kwarg parsing in C API

--56738661_23f9c13c_16c
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Joseph Jones liked your message with Boxer. On December 1, 2015 at 19:23:=
44 MST, Eric Wong <normalperson=40yhbt.net> wrote:normalperson=40yhbt.net=
 wrote:> https://bugs.ruby-lang.org/issues/11339=46or OpenSSL connect=5Fn=
onblock/accept=5Fnonblock, it seems to be not worththe effort for a 1% im=
provement given the overheads of various parts ofOpenSSL. But I'm also no=
t knowledgeable in OpenSSL, so my benchmark isalso likely bogus: http://8=
0x24.org/spew/20151202020654.18328-1-e%4080x24.org/Additionally, for read=
=5Fnonblock; calling read=5Fnonblock/sysreadvia rb=5Ffuncall from inside =
ossl=5Fssl=5Fread=5Finternal would also requiremore work to avoid the has=
h allocation. Not sure if it's worththe effort at the moment.     
--56738661_23f9c13c_16c
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<html><body><div>Joseph Jones liked your message with <a href=3D=22http:/=
/bxr.io/PBIqo=22>Boxer</a>.</div><br/><br/><div><div class=3D=22quote=22>=
On December 1, 2015 at 19:23:44 MST, Eric Wong &lt;normalperson=40yhbt.ne=
t&gt; wrote:<br/><blockquote type=3D=22cite=22 style=3D=22border-left-sty=
le:solid;border-width:1px;margin-left:0px;padding-left:10px;=22>normalper=
son=40yhbt.net wrote:<br />> https://bugs.ruby-lang.org/issues/11339<br /=
><br />=46or OpenSSL connect=5Fnonblock/accept=5Fnonblock, it seems to be=
 not worth<br />the effort for a 1% improvement given the overheads of va=
rious parts of<br />OpenSSL.  But I'm also not knowledgeable in OpenSSL, =
so my benchmark is<br />also likely bogus:<br /><br />  http://80x24.org/=
spew/20151202020654.18328-1-e%4080x24.org/<br /><br />Additionally, for r=
ead=5Fnonblock; calling read=5Fnonblock/sysread<br />via rb=5Ffuncall fro=
m inside ossl=5Fssl=5Fread=5Finternal would also require<br />more work t=
o avoid the hash allocation.  Not sure if it's worth<br />the effort at t=
he moment.<br /></blockquote></div></div></body></html>
--56738661_23f9c13c_16c--