**** BEGIN LOGGING AT Thu Apr 25 02:59:59 2013
Apr 25 07:39:09 <apelete> Good morning
Apr 25 08:39:01 <hrw> hello
Apr 25 08:50:55 <apelete> Hi hrw
Apr 25 09:09:53 <pb_> hi all
Apr 25 09:14:30 <florian> morning pb_ 
Apr 25 09:30:02 <apelete> hi pb_ 
Apr 25 09:39:24 <silviof> morning all
Apr 25 10:34:36 <Crofton|work> hmm, I've discovered a new use for OE, hunting for 32 bit x86 compile fails
Apr 25 10:46:41 <hrw> Crofton|work: I still miss one dependency to finally make 'bitbake pizza'
Apr 25 10:47:33 <Crofton|work> we should work on that
Apr 25 10:48:15 <hrw> Crofton|work: I only need one of pizza places near me to start accepting orders by internet
Apr 25 10:48:33 <hrw> then it will be a matter of recipe + custom do_fetch
Apr 25 10:49:06 <hrw> useless in a way RFC1149 implementation was useless
Apr 25 10:55:31 <Crofton|work> we shuold make an entry in bugzilla
Apr 25 11:10:58 * tsjsieb agrees, how high can the priority be in bugzilla?
Apr 25 11:11:59 <tsjsieb> although I could live with a 19" server with special pizza slot
Apr 25 12:41:24 <hsychla> hi guys, quick basic question: how do i check in python is a variable is unset and if so, set it? I tried base_conditional but it won't work like CORE_NUMBER = "${@base_conditional(´CORE_NUMBER´, ´´, ´1´, ´${CORE_NUMBER}´, d)}" ? any tips?
Apr 25 14:35:02 <Crofton|work> I need to fix numpy for x86_64
Apr 25 14:35:17 <Crofton|work> what is the directory name /override for 64 bit builds
Apr 25 14:35:24 <Crofton|work> looks like i586 for 32 bit
Apr 25 15:01:37 <zecke> pb_: OT: I wonder if you can give me a hint/solution for a performance analysis. :}
Apr 25 15:01:46 <pb_> zecke: I can try
Apr 25 15:03:15 <zecke> pb_: I have an interpreter using computed goto and all opcodes are inlined and in one C function. Do you know an asm("") trick to make this show up as one function per symbol? To be honest I haven't tried any inline assembly yet. :)
Apr 25 15:05:12 <pb_> well, you can certainly inject random symbols into the output using asm().  You'd need to make them volatile to stop the compiler moving code across them, and that of course might inhibit some optimisations.
Apr 25 15:05:23 <pb_> What are you trying to do exactly?
Apr 25 15:08:06 <zecke> pb_: The high level goal is to make my interpreter run more quickly. This software is quite optimized by itself so I would like to make sure that opcodes used a lot sit as close as possible together. So being able to see different symbols in perf would be nice
Apr 25 15:11:16 <pb_> right, I see.  I guess the complicating factor is that (in general) there is no guarantee that the code for a given opcode will be contiguous, though in practice your interpreter routines might be small enough that they always are.
Apr 25 15:11:47 <pb_> You could certainly try just emitting some symbols and see what happens, asm volatile ("zecke_is_cool:") or something
Apr 25 15:12:36 <zecke> yeah. i did this and to my surprise that already worked i thought i would have to write .func and .cti..
Apr 25 15:12:39 <zecke> thanks
Apr 25 15:19:46 <zecke> okay but perf gets fooled by the computed goto.. "jump behind the symbol"
Apr 25 19:07:07 <pb_> evening all
Apr 25 19:10:36 <woglinde> hi pb
Apr 25 19:38:53 <pb_> hi woglinde
**** ENDING LOGGING AT Fri Apr 26 02:59:58 2013