Comments on: A more in-depth analysis of Ruby HTTP client performance http://apocryph.org/2008/11/09/more_indepth_analysis_ruby_http_client_performance/ Notes to my future self Tue, 09 Mar 2010 17:12:01 -0600 http://wordpress.org/?v=2.8.6 hourly 1 By: 7fff - think max value » Blog Archive » Faster Net::HTTP for Ruby 1.8.6 http://apocryph.org/2008/11/09/more_indepth_analysis_ruby_http_client_performance/comment-page-1/#comment-251 7fff - think max value » Blog Archive » Faster Net::HTTP for Ruby 1.8.6 Sat, 20 Dec 2008 15:11:46 +0000 http://apocryph.org/?p=593#comment-251 [...] been a bit frustrated at work with Net::HTTP performance (as have so many others) so here’s a monkeypatch for 1.8.6 that combines the buffer size increase in 1.8.7 with Aaron [...] [...] been a bit frustrated at work with Net::HTTP performance (as have so many others) so here’s a monkeypatch for 1.8.6 that combines the buffer size increase in 1.8.7 with Aaron [...]

]]> By: anelson http://apocryph.org/2008/11/09/more_indepth_analysis_ruby_http_client_performance/comment-page-1/#comment-218 anelson Mon, 10 Nov 2008 23:37:34 +0000 http://apocryph.org/?p=593#comment-218 I didn't think so, but I just checked and sure enough, there is a binary gem for version 0.12.0. Thanks for pointing that out. I didn’t think so, but I just checked and sure enough, there is a binary gem for version 0.12.0. Thanks for pointing that out.

]]> By: Abdul-Rahman Advany http://apocryph.org/2008/11/09/more_indepth_analysis_ruby_http_client_performance/comment-page-1/#comment-219 Abdul-Rahman Advany Mon, 10 Nov 2008 22:40:16 +0000 http://apocryph.org/?p=593#comment-219 is it? I thought eventmachine had binary releases for windows... is it? I thought eventmachine had binary releases for windows…

]]> By: anelson http://apocryph.org/2008/11/09/more_indepth_analysis_ruby_http_client_performance/comment-page-1/#comment-220 anelson Mon, 10 Nov 2008 15:08:09 +0000 http://apocryph.org/?p=593#comment-220 A flurry of warnings, followed by: In file included from rev_loop.c:9: /home/anelson/ruby19/include/ruby-1.9.0/ruby/backward/rubysig.h:14:2: warning: #warning rubysig.h is obsolete In file included from rev.h:11, from rev_loop.c:14: /home/anelson/ruby19/include/ruby-1.9.0/ruby/backward/rubyio.h:2:2: warning: #warning use "ruby/io.h" instead of "rubyio.h" rev_loop.c: In function 'Rev_Loop_ev_loop_oneshot': rev_loop.c:211: error: 'RB_UBF_DFL' undeclared (first use in this function) rev_loop.c:211: error: (Each undeclared identifier is reported only once rev_loop.c:211: error: for each function it appears in.) rev_loop.c:287:2: warning: no newline at end of file make: *** [rev_loop.o] Error 1 I admit I didn't spend a whole lot of time looking into it, though I at least got as far as getting libev installed. A flurry of warnings, followed by:

In file included from rev_loop.c:9:
/home/anelson/ruby19/include/ruby-1.9.0/ruby/backward/rubysig.h:14:2: warning: #warning rubysig.h is obsolete
In file included from rev.h:11,
from rev_loop.c:14:
/home/anelson/ruby19/include/ruby-1.9.0/ruby/backward/rubyio.h:2:2: warning: #warning use “ruby/io.h” instead of “rubyio.h”
rev_loop.c: In function ‘Rev_Loop_ev_loop_oneshot’:
rev_loop.c:211: error: ‘RB_UBF_DFL’ undeclared (first use in this function)
rev_loop.c:211: error: (Each undeclared identifier is reported only once
rev_loop.c:211: error: for each function it appears in.)
rev_loop.c:287:2: warning: no newline at end of file
make: *** [rev_loop.o] Error 1

I admit I didn’t spend a whole lot of time looking into it, though I at least got as far as getting libev installed.

]]> By: Tony Arcieri http://apocryph.org/2008/11/09/more_indepth_analysis_ruby_http_client_performance/comment-page-1/#comment-222 Tony Arcieri Mon, 10 Nov 2008 05:25:40 +0000 http://apocryph.org/?p=593#comment-222 What problems did you encounter with Rev? What problems did you encounter with Rev?

]]> By: anelson http://apocryph.org/2008/11/09/more_indepth_analysis_ruby_http_client_performance/comment-page-1/#comment-223 anelson Mon, 10 Nov 2008 04:09:10 +0000 http://apocryph.org/?p=593#comment-223 Indeed. The more I learn how shitty the stock 1.8.6 HTTP client impl is, the more surprised I am that I haven't read more about it. It's basically useless if you're doing any sort of HTTP interface on the server side, particularly with miserly shared hosts. Really lame. Indeed. The more I learn how shitty the stock 1.8.6 HTTP client impl is, the more surprised I am that I haven’t read more about it. It’s basically useless if you’re doing any sort of HTTP interface on the server side, particularly with miserly shared hosts. Really lame.

]]> By: Cliff Moon http://apocryph.org/2008/11/09/more_indepth_analysis_ruby_http_client_performance/comment-page-1/#comment-221 Cliff Moon Mon, 10 Nov 2008 03:57:56 +0000 http://apocryph.org/?p=593#comment-221 We saw the same issues getting ready for launch at Powerset. The frontend talks to a lot of backend services, some of which run over http. CPU time for http transfer was saturating the machines. As a stopgap measure I monkey-patched Net::HTTP so that it had a larger buffer and didn't do a timeout every read, at least until we could port all the code over to use libcurl. We saw the same issues getting ready for launch at Powerset. The frontend talks to a lot of backend services, some of which run over http. CPU time for http transfer was saturating the machines. As a stopgap measure I monkey-patched Net::HTTP so that it had a larger buffer and didn’t do a timeout every read, at least until we could port all the code over to use libcurl.

]]>