Text of wiki.python.org › python › networking › NeedForSpeed › IRCLog.html as Styx crawled it on Nov 19, 2026. The page may have changed since. Only its words are kept, without pictures or layout, and only the first part of a long page.

Open the live page

Need for Speed / IRC Log - Python Wiki

Python Wiki Python Wiki / Python Wiki / Networking / Needforspeed / Need for Speed / IRC Log View as Markdown Need for Speed / IRC Log ¶ Legacy Wiki Page This page was migrated from the old MoinMoin-based wiki. Information may be outdated or no longer applicable. For current documentation, see python.org . 2006-05-22 05:10:40> *** nfsbot has joined #nfs 2006-05-22 05:10:40> *** nfsbot has joined #nfs 2006-05-22 05:11:08> <jafo> Hello nfsbot.

2006-05-22 05:12:00> <jafo> Ok, does anyone want the IRC logs on the web live as they are happening, or can I just wait until we're done and publish then? 2006-05-22 05:13:41> *** blais has joined #nfs 2006-05-22 05:14:20> <rjones> whatever :) 2006-05-22 05:15:31> *** dalke has joined #nfs 2006-05-22 05:15:44> <jafo> dalke: Hey there. 2006-05-22 05:19:42> <effbot> anyone here? 2006-05-22 05:19:59> <effbot> have everyone found anything to do, or ...

2006-05-22 05:23:09> <jafo> I'm just building my task list now. 2006-05-22 05:26:33> *** jbenedik has quit IRC 2006-05-22 05:39:00> *** jbenedik has joined #nfs 2006-05-22 05:41:53> *** ymmit has quit IRC 2006-05-22 05:41:59> <jafo> Is anyone working on trying to convert to 64-bit ints on 32-bit platforms and see how it impacts pybench? 2006-05-22 05:45:35> <jafo> holdenweb_: I still think we need to have a group discussion on the slow-down from 2.4 to 2.5 that you saw.

2006-05-22 05:48:34> *** blais has quit IRC 2006-05-22 05:56:51> *** holdenweb_ has quit IRC 2006-05-22 06:03:40> *** jbenedik_ has joined #nfs 2006-05-22 06:06:36> *** runarp has joined #nfs 2006-05-22 06:07:00> *** kristjan_ has joined #nfs 2006-05-22 06:07:13> *** ymmit has joined #nfs 2006-05-22 06:07:35> *** sholden__ has joined #nfs 2006-05-22 06:08:22> *** etrepum_ has joined #nfs 2006-05-22 06:09:45> *** holdenweb_ has joined #NFS 2006-05-22 06:11:54> *** jbenedik_ has quit IRC 2006-05-22 06:15:44> *** gbrandl has quit IRC 2006-05-22 06:15:49> *** jbenedik has quit IRC 2006-05-22 06:15:52> *** rjones has quit IRC 2006-05-22 06:15:52> *** grunar has quit IRC 2006-05-22 06:16:05> *** etrepum has quit IRC 2006-05-22 06:16:08> *** gbrandl has joined #nfs 2006-05-22 06:16:14> *** sholden_ has quit IRC 2006-05-22 06:16:18> *** sholden has quit IRC 2006-05-22 06:16:22> *** dalke has quit IRC 2006-05-22 06:16:39> *** kristjan has quit IRC 2006-05-22 06:16:47> *** sholden has joined #nfs 2006-05-22 06:20:17> *** effbot has quit IRC 2006-05-22 06:21:32> *** uncletimmy has joined #nfs 2006-05-22 06:21:43> *** effbot has joined #nfs 2006-05-22 06:21:56> <effbot> I've posted my slowdown results here: 2006-05-22 06:21:57> <effbot> http://wiki.python.org/moin/NeedForSpeed/Goals/Slowdown#preview 2006-05-22 06:22:26> <effbot> The big ones seem to be import and try/except.

2006-05-22 06:22:28> <jafo> effbot: I'm just running tests now. 2006-05-22 06:25:17> *** jack_diederich has joined #nfs 2006-05-22 06:27:08> <holdenweb_> anyone using svn tortoise can show me how to set up authentication? 2006-05-22 06:32:15> *** ymmit has quit IRC 2006-05-22 06:40:30> *** etrepum has joined #nfs 2006-05-22 06:40:52> *** gbr_ has joined #nfs 2006-05-22 06:41:25> *** kristjan has joined #nfs 2006-05-22 06:42:10> *** holdenweb has joined #NFS 2006-05-22 06:45:36> *** runarp has quit IRC 2006-05-22 06:55:56> *** uncletimmy has quit IRC 2006-05-22 06:58:36> *** jack_diederich has quit IRC 2006-05-22 06:58:39> *** etrepum_ has quit IRC 2006-05-22 06:58:43> *** sholden_ has joined #nfs 2006-05-22 06:58:58> *** sholden__ has quit IRC 2006-05-22 06:58:58> *** jack_diederich has joined #nfs 2006-05-22 06:59:00> *** sholden has quit IRC 2006-05-22 06:59:11> *** kristjan_ has quit IRC 2006-05-22 06:59:20> *** sholden has joined #nfs 2006-05-22 06:59:42> *** effbot has quit IRC 2006-05-22 07:00:24> *** gbrandl has quit IRC 2006-05-22 07:01:36> *** holdenweb_ has quit IRC 2006-05-22 07:39:38> *** effbot has joined #nfs 2006-05-22 07:44:50> *** grunar has joined #nfs 2006-05-22 07:56:08> *** runarp has joined #nfs 2006-05-22 08:00:10> *** rxe has joined #nfs 2006-05-22 08:01:14> <effbot> I have gobby running on 192.168.0.103 port 6522 if anyone wants to try.

2006-05-22 08:11:05> <holdenweb> effbot: gobby 0.3 or 0.4? 2006-05-22 08:14:54> *** grunar has quit IRC 2006-05-22 08:20:08> <effbot> gobby 0.4 2006-05-22 08:36:52> <holdenweb> so i discovered when I ran 0.3 :_) 2006-05-22 08:43:48> *** kristjan_ has joined #nfs 2006-05-22 08:47:06> *** ymmit has joined #nfs 2006-05-22 08:51:15> <holdenweb> effbot: how come I get an "overhead" column on my pybench output where you have a "diff *)" column?

2006-05-22 09:02:49> *** kristjan has quit IRC 2006-05-22 09:13:36> *** jbenedik has joined #nfs 2006-05-22 09:17:59> *** blais has joined #nfs 2006-05-22 09:19:35> *** ymmit has quit IRC 2006-05-22 09:23:37> *** ymmit has joined #nfs 2006-05-22 09:24:11> <jafo> I'm going to take a look at switching ints to 64 bits on 32-bit platforms. 2006-05-22 09:42:40> *** blais has quit IRC 2006-05-22 09:42:51> *** blais has joined #nfs 2006-05-22 09:46:49> *** b-_-d has joined #nfs 2006-05-22 09:47:04> <b-_-d> anyone use nfs? how do i get locking?

2006-05-22 09:54:15> <jafo> b-_-d: See the channel topic. 2006-05-22 09:58:57> *** kristjan_ has quit IRC 2006-05-22 10:06:51> <jafo> How does Python decide to do the up-convert from int to long? I've converted all the types, I think, to "long long", but 1<<32 is going to L. 2006-05-22 10:12:02> *** doctorwells has joined #nfs 2006-05-22 10:13:09> <ymmit> jafo: for 1<<32, see intobject.c's int_lshift. 2006-05-22 10:13:20> <jafo> Thanks.

2006-05-22 10:13:57> <ymmit> You're probably see that the platform LONG_BIT is #define'd as 32, so left shift figures it _needs_ a Python long. 2006-05-22 10:14:35> <b-_-d> channel name is deciving 2006-05-22 10:14:59> <b-_-d> #Python would work 2006-05-22 10:15:10> *** b-_-d has left #nfs 2006-05-22 10:15:11> <jafo> b-_-d: No, it wouldn't. 2006-05-22 10:30:48> <jbenedik> doctorwells: you at keys? 2006-05-22 10:31:57> <doctorwells> yup 2006-05-22 10:34:21> *** uncletimmy has joined #nfs 2006-05-22 10:39:37> *** ymmit has quit IRC 2006-05-22 10:41:05> <jbenedik> i'm working on psyco dicts right now 2006-05-22 10:44:14> <doctorwells> cool, what part? virtualized keys(), values() and items() ?

2006-05-22 10:44:37> <jbenedik> we'll see how much - ideally yes 2006-05-22 10:44:40> <jbenedik> faster iteration 2006-05-22 10:44:44> <jbenedik> virtualized access 2006-05-22 10:46:11> <jafo> Ok, this is extremely weird. I converted the int stuff to long long on 32-bit platforms, and I'm getting: 2006-05-22 10:46:13> <jafo> Average round time: 4081.70 ms -100.00% 2006-05-22 10:46:26> <jafo> The -100.00% is consistent.

2006-05-22 10:47:02> <doctorwells> sounds a lot faster 2006-05-22 10:48:08> <jafo> But the original is: Average round time: 4037.00 ms 2006-05-22 10:48:25> <jafo> Oh, wait, obviously there's a problem in my conversion to long long... 2006-05-22 10:48:34> <rxe> doctorwells: joining the sprint remotely? :-) 2006-05-22 10:48:44> <doctorwells> sumthin' like that 2006-05-22 10:50:19> <jafo> So, it looks like on pybench the difference is 1.11% slow-down by going to long long.

2006-05-22 10:51:03> <jafo> So, do we want to suck it up, or give it a pass? Anyone? 2006-05-22 10:51:26> <jbenedik> What kind of speedup do you get when you overflow? any? 2006-05-22 10:51:40> <doctorwells> does pybench do anything with ints in the 32 - 64 bit range? 2006-05-22 10:51:53> <jafo> Don't know and don't know. 2006-05-22 10:55:17> <jafo> Looks like 1<<35 is around 34% faster. 2006-05-22 11:13:24> *** runarp has quit IRC 2006-05-22 11:13:24> *** jafo has quit IRC 2006-05-22 11:13:30> *** runarp has joined #nfs 2006-05-22 11:13:30> *** jafo has joined #nfs 2006-05-22 11:37:12> *** kristjan has joined #nfs 2006-05-22 11:49:22> <kristjan> CS:EIP Symbol + Offset Thread ID Samples Total % CPU0 32 bit 64 bit 2006-05-22 11:49:22> <kristjan> 0x1e00f070 PyEval_EvalFrameEx 12839 49.94 12839 12839 0 2006-05-22 11:49:22

The rest of this page is on the live site.