4 ms·
A second long sleep on every read or write? If this was actually happening, it sounds like it could create unheard of performance issues for any significant tr
by timf 16y ago
A second long sleep on every read or write? If this was actually happening, it sounds like it could create unheard of performance issues for any significant transfer.
Or was this not noticed because all the major frameworks like cherrypy and twisted are still using the pyopenssl wrapper?
Is there any evidence that this bugfix actually changes the performance?
- MrRage 16y agoDon't you mean a millisecond long sleep? I never saw a sleep function that interpreted its parameter as seconds. Edit: Well, now I know better.
- kingkilr 16y agoIt does on Linux, from `man 3 sleep`: http://paste.pocoo.org/show/229678/ http://paste.pocoo.org/show/229678/
- timf 16y agoThis is what I was going on, I checked that man page, incredulously...
- dmm 16y agoBoth the POSIX sleep() program and the sleep(unsigned int seconds) function in unistd.h interpret the argument as seconds.