| 02:10.21 | *** join/#brlcad dli (n=dli@wireless-194-151.uchicago.edu) |
| 03:13.33 | louipc | ew |
| 06:14.04 | CIA-7 | BRL-CAD: 03brlcad * 10brlcad/BUGS: |
| 06:14.04 | CIA-7 | BRL-CAD: add at least three (rather minor, all with workarounds) bugs noticed relating to |
| 06:14.04 | CIA-7 | BRL-CAD: libfb/ray-tracer file redirection. libfb cannot utilize standard output/error |
| 06:14.04 | CIA-7 | BRL-CAD: 'device' files (e.g. /dev/stdout) without path trickery, using such a device |
| 06:14.04 | CIA-7 | BRL-CAD: fails if used with a pipe or redirect, and the ray-tracers output corrupted |
| 06:14.07 | CIA-7 | BRL-CAD: imagery if using -o to a non-seekable file (like /dev/stdout). |
| 07:00.14 | *** join/#brlcad clock_ (i=clock@84-72-90-161.dclient.hispeed.ch) |
| 08:43.48 | *** join/#brlcad clock_ (n=clock@zux221-122-143.adsl.green.ch) |
| 09:31.55 | *** join/#brlcad b0ef (n=b0ef@084202026226.customer.alfanett.no) |
| 13:53.25 | ``Erik | *yawn* |
| 14:54.02 | Maloeran | Grah, I'm really bad sometimes. Struggling a moment with merely solving boolean operations on 2d polygons |
| 15:29.15 | *** join/#brlcad IriX64 (n=mario_du@bas3-sudbury98-1168049030.dsl.bell.ca) |
| 15:39.21 | IriX64 | brlcad: I've abandoned that convertor thing you asked me to do, don't know enough about the various formats, and lack test facilities for those, easiest way for you ppl though is convert one way then convert back to where you started from, should get an exact copy of what you started with. |
| 16:24.38 | IriX64 | now to learn how to interact with adrt/rise. |
| 17:31.17 | *** join/#brlcad elzair (n=elzair@csdhcp88.cs.clemson.edu) |
| 17:44.39 | ``Erik | heh |
| 17:47.33 | ``Erik | g-dxf -o moss.dxf moss.g all.g && dxf-g moss.dxf moss-new.g && mged moss-new.g .... then e up the top level and see if it looks right... |
| 17:47.40 | ``Erik | don't really need to know much about the formats |
| 18:05.13 | CIA-7 | BRL-CAD: 03brlcad * 10brlcad/bench/run.sh: pixcmp now reports non-zero code for off-by-one results. handle the new/expected return values. |
| 19:51.04 | IriX64 | ``Erik, i don't speak nix cmd lin :) |
| 20:19.20 | brlcad | wicked |
| 20:19.46 | brlcad | turning off debugging symbols actually decreased performance by a solid amount |
| 20:20.55 | brlcad | 25%! |
| 20:21.27 | brlcad | that's gotta be some massive alignment/coherency penalty going on |
| 20:22.00 | dtidrow_work | talk about bizarre |
| 20:24.20 | brlcad | yeah, both O3, exact same options .. just one with -g and one without |
| 20:24.49 | brlcad | adding alignment flags to see |
| 20:28.32 | ``Erik | brlcad, what am I doing wrong or stupidly? error: cannot find interface declaration for 'NSObject', superclass of 'NSAppleMenuController |
| 20:29.15 | brlcad | #include <CoreFoundation/CoreFoundation.h> |
| 20:29.55 | ``Erik | still get it :/ |
| 20:30.29 | brlcad | er, one of those |
| 20:30.30 | ``Erik | even added "-framework CoreFoundation" to the compile line to see if it fixes something, but *shrug* no joy |
| 20:30.34 | brlcad | try Foundation/Foundation |
| 20:31.14 | *** join/#brlcad docelic (n=docelic@212.15.185.220) |
| 20:33.13 | ``Erik | hm, Foundation seems to be fixing some of it |
| 20:38.20 | ``Erik | yeah, heh |
| 20:38.34 | ``Erik | this .m file is... not... healthy... |
| 20:38.48 | ``Erik | at least, not for the twisted way I'm trying to abuse it O:-) |
| 20:38.51 | ``Erik | thanks :D |
| 20:39.18 | brlcad | SDLMain.m? |
| 20:39.34 | ``Erik | yeah |
| 20:39.36 | ``Erik | ahhhhhh |
| 20:39.43 | ``Erik | #import <Cocoa/Cocoa.h> made it all happy |
| 20:40.08 | brlcad | heh, including the kitchen sink usually does |
| 20:41.14 | brlcad | you do realize that there's an SDLMain.h too? |
| 20:41.34 | brlcad | that includes SDL.h, Cocoa.h, and declares the interface correctly |
| 20:43.45 | ``Erik | ah, heh, didn't notice it... the .m didn't import it :/ |
| 21:12.53 | brlcad | ahh, function/jump/loop alignment alone restored performance back to within about 95% .. i bet it'd be all back if I added a few more related flags |
| 21:13.52 | brlcad | do we have any popular systems that aren't aligned on 16 other than hpc? |
| 21:14.40 | *** join/#brlcad clock_ (i=clock@84-72-60-125.dclient.hispeed.ch) |
| 21:24.14 | *** join/#brlcad Elperion (n=Elperion@p54876260.dip.t-dialin.net) |
| 21:35.19 | Maloeran | Very often, chunks of memory from malloc() are only aligned on 8 bytes |
| 22:05.50 | *** join/#brlcad louipc (n=louipc@bas8-toronto63-1096669545.dsl.bell.ca) |
| 23:54.50 | ``Erik | .. that depends on OS |