| 00:13.24 |
CIA-77 |
BRL-CAD: 03starseeker * r43530
10/brlcad/branches/cmake/CMakeLists.txt: Remove the complex and
only partially successful noprod logic - with targets in toplevel
bin dirs anyway the utility is minimal, and not worth the
complexity. |
| 00:36.45 |
CIA-77 |
BRL-CAD: 03starseeker * r43531
10/brlcad/branches/cmake/ (3 files in 3 dirs): Ignore other in
src |
| 00:39.42 |
CIA-77 |
BRL-CAD: 03starseeker * r43532
10/brlcad/branches/cmake/src/librt/CMakeLists.txt: Ignore
search.h |
| 00:50.20 |
*** join/#brlcad Klebel
(~mk@w73.RIC.Berkeley.EDU) |
| 00:50.49 |
Klebel |
I can't find 'Set H' in the Edit
menu |
| 00:57.36 |
Klebel |
press "Set H" - says unknown
operation. |
| 00:57.45 |
Klebel |
on the command line |
| 00:58.31 |
starseeker |
Klebel: we need more context |
| 00:59.29 |
Klebel |
page 60 in the Introduction to MGED
manual. |
| 00:59.29 |
Klebel |
pdf |
| 01:00.22 |
Klebel |
I created a right circular cylinder, then it
tells me to do, "Edit and then Set H" |
| 01:00.33 |
Klebel |
and click the middle mouse button several
times |
| 01:01.02 |
starseeker |
to edit a primitive, you need to use
sed |
| 01:01.08 |
starseeker |
that puts you in edit mode |
| 01:02.05 |
starseeker |
if you aren't seeing Set H you probably aren't
in edit mode |
| 01:02.08 |
Klebel |
ah, so sed base1.s |
| 01:02.08 |
Klebel |
thanks, Set H is now in the menu, and
works |
| 01:02.29 |
starseeker |
that tutorial was created with an older
version of BRL-CAD, so there are occasional differences |
| 01:02.43 |
Klebel |
yea I've noticed that :/ |
| 01:07.15 |
Klebel |
is there a command to get out of edit
mode? |
| 01:07.26 |
starseeker |
accept or reject |
| 01:07.40 |
Klebel |
ok thanks |
| 01:09.02 |
CIA-77 |
BRL-CAD: 03starseeker * r43533
10/brlcad/branches/cmake/ (27 files in 22 dirs): MFC
r43532 |
| 01:10.42 |
CIA-77 |
BRL-CAD: 03starseeker * r43534
10/brlcad/branches/cmake/src/other/CMakeLists.txt: Uncomment -w
again for src/other |
| 01:17.39 |
CIA-77 |
BRL-CAD: 03starseeker * r43535
10/brlcad/branches/cmake/misc/CMake/CompilerFlags.cmake: We're
going for gnu99 now |
| 02:01.02 |
CIA-77 |
BRL-CAD: 03brlcad * r43536
10/brlcad/trunk/src/bwish/cadAppInit.c: include bin.h instead of
winsock2.h |
| 03:28.05 |
*** join/#brlcad guest_tttt
(~rm@123.136.11.66) |
| 03:34.43 |
*** part/#brlcad guest_tttt
(~rm@123.136.11.66) |
| 04:32.23 |
*** join/#brlcad epileg
(~epileg@unaffiliated/epileg) |
| 04:33.44 |
CIA-77 |
BRL-CAD: 03brlcad * r43537
10/brlcad/trunk/src/rttherm/pixtest.c: check fwrite return
value |
| 04:36.11 |
CIA-77 |
BRL-CAD: 03brlcad * r43538
10/brlcad/trunk/src/sig/ (24 files): check fwrite return values for
failure |
| 04:39.55 |
CIA-77 |
BRL-CAD: 03brlcad * r43539
10/brlcad/trunk/src/proc-db/brepintersect.cpp: compiler is
complaining about the first param to SegmentPolylineIntersect
possibly being NULL. as this is dev code, just comment out for now
in leu of removing the code. |
| 04:44.14 |
CIA-77 |
BRL-CAD: 03brlcad * r43540
10/brlcad/trunk/src/mged/points/points_parse.y: bison is being
stupid with some output code generating size_t comparisons against
>= 0. quell that warning along with a couple other preprocessor
symbols that are not defined but being used in
expressions. |
| 04:45.52 |
CIA-77 |
BRL-CAD: 03brlcad * r43541
10/brlcad/trunk/src/ (mged/points/points_scan.l tab/script.l):
quell flex lameness where fwrite() is being called without checking
the return value. this quiets the compiler. |
| 04:51.08 |
CIA-77 |
BRL-CAD: 03brlcad * r43542
10/brlcad/trunk/src/ (23 files in 7 dirs): |
| 04:51.09 |
CIA-77 |
BRL-CAD: categorically check return values for
some of the stdio and stdlib routines |
| 04:51.09 |
CIA-77 |
BRL-CAD: (e.g. fwrite, scanf, system, dup,
pipe, ...). not willing to put forth |
| 04:51.09 |
CIA-77 |
BRL-CAD: time/effort to do anything more than
print the error since would have to |
| 04:51.09 |
CIA-77 |
BRL-CAD: evaluate each call on a case by case
basis (and that's not fun). |
| 05:10.50 |
*** join/#brlcad dli
(~dli@dsl-67-204-45-87.acanac.net) |
| 05:35.23 |
*** join/#brlcad Stattrav
(~Stattrav@122.167.254.137) |
| 05:35.23 |
*** join/#brlcad Stattrav
(~Stattrav@unaffiliated/stattrav) |
| 06:08.23 |
CIA-77 |
BRL-CAD: 03brlcad * r43543
10/brlcad/trunk/src/libfb/if_X24.c: fix keybindings on Mac OS X so
that cmd-click will produce button 3 events (so we can close
framebuffers) without needing the user to set X11.app to emulate a
three button mouse. |
| 06:09.28 |
CIA-77 |
BRL-CAD: 03brlcad * r43544
10/brlcad/trunk/src/libfb/if_ogl.c: do the same mouse-3 binding for
ogl |
| 06:51.11 |
Klebel |
1 thing I keep noticing in the Introduction
manual, is that when I copy primatives through the command line
then run sed on the newly copied primative it says: Error sph2.s
not being displayed |
| 06:51.37 |
Klebel |
the only way I am able to copy is through the
Primitive Editor |
| 07:01.04 |
CIA-77 |
BRL-CAD: 03brlcad * r43545
10/brlcad/trunk/src/tclscripts/mged/bindings.tcl: |
| 07:01.04 |
CIA-77 |
BRL-CAD: fix mged zoom bindings on mac os x
with default X11 (where 3-button mouse |
| 07:01.04 |
CIA-77 |
BRL-CAD: emulation is disabled) so that you
can actually zoom out. make cmd-click behave |
| 07:01.04 |
CIA-77 |
BRL-CAD: the same as button-3. unfortunately,
the same binding does not seem possible |
| 07:01.04 |
CIA-77 |
BRL-CAD: with option-click for mouse 2 (in
fact, option seems to act like mod2 aka the |
| 07:01.04 |
CIA-77 |
BRL-CAD: command-key. none of the other mod
types seem to help. |
| 07:07.37 |
CIA-77 |
BRL-CAD: 03brlcad * r43546
10/brlcad/trunk/NEWS: |
| 07:07.38 |
CIA-77 |
BRL-CAD: fixed a problem being able to zoom in
with the default x11.app configuration |
| 07:07.38 |
CIA-77 |
BRL-CAD: where you could zoom out, but not
back in without enabling 3-button emulation. |
| 07:07.38 |
CIA-77 |
BRL-CAD: this binds command+button1 the same
as button3. couldn't get option+button1 to |
| 07:07.38 |
CIA-77 |
BRL-CAD: behave as button2 though. |
| 07:10.21 |
CIA-77 |
BRL-CAD: 03brlcad * r43547
10/brlcad/trunk/src/librt/primitives/table.c: tested and
rt_generic_xform() is NOT sufficient. wrong plot and trace with
loads of error. |
| 07:10.47 |
brlcad |
Klebel: when you copy something, it's not
automatically drawn |
| 07:10.57 |
brlcad |
cp old new ; e new |
| 07:11.07 |
brlcad |
then sed and friends will work |
| 07:11.12 |
brlcad |
(on new) |
| 07:11.25 |
brlcad |
e == draw |
| 07:13.30 |
brlcad |
starseeker: mged regression is failing with
"Tcl_Import ERROR: unknown namespace in import pattern
"::itcl::*" |
| 07:14.09 |
brlcad |
wondering if you're seeing that on your
end |
| 07:14.50 |
brlcad |
only namespace or itcl change that comes to
mind is the mged/bwish setup routine |
| 07:15.47 |
brlcad |
yeah, namespace import fails because itcl_init
is failing, can't find/load the itcl .so file (testing on linux
atm) |
| 07:52.39 |
Klebel |
thanks brlcad |
| 10:18.24 |
*** join/#brlcad epileg
(~epileg@unaffiliated/epileg) |
| 10:38.35 |
*** join/#brlcad Elrohir
(~kvirc@p5B149820.dip.t-dialin.net) |
| 11:07.56 |
*** join/#brlcad Elrohir
(~kvirc@p5B149820.dip.t-dialin.net) |
| 11:13.53 |
*** join/#brlcad Stattrav
(~Stattrav@unaffiliated/stattrav) |
| 11:30.56 |
dloman |
Mernin all. |
| 11:46.40 |
CIA-77 |
BRL-CAD: 03d_rossberg * r43548
10/brlcad/trunk/include/bin.h: |
| 11:46.40 |
CIA-77 |
BRL-CAD: undef some windows defines which are
used in an other context here |
| 11:46.40 |
CIA-77 |
BRL-CAD: (same as in bio.h) |
| 11:49.56 |
CIA-77 |
BRL-CAD: 03d_rossberg * r43549
10/brlcad/trunk/ (3 files in 3 dirs): raytrace.h got a regex.h
include -- added the corresponding search path |
| 12:14.00 |
starseeker |
brlcad: I'll have to check - I'm doing most of
my development work with CMake these days, so I haven't tried the
autotools regression in a while |
| 12:14.12 |
starseeker |
I thought it worked, but maybe it was a
fluke... |
| 12:27.29 |
dloman |
Huh. Neato:
http://sewelldirect.com/Sewell-Minideck-USB-to-DVI-Display-Adapter.asp |
| 12:45.21 |
CIA-77 |
BRL-CAD: 03starseeker * r43550
10/brlcad/trunk/src/fbserv/fbserv.c: Shadowing a global -
fixed |
| 12:47.24 |
CIA-77 |
BRL-CAD: 03starseeker * r43551
10/brlcad/trunk/src/lgt/ (hmenu.c lgt.c): Remove defined-but-unused
functions causing lgt failures. |
| 12:49.29 |
starseeker |
brlcad: still getting a script.c
failure: |
| 12:49.40 |
starseeker |
script.c: In function
âyy_get_next_bufferâ: |
| 12:49.40 |
starseeker |
script.c:1414: error: comparison between
signed and unsigned integer expressions |
| 13:07.12 |
starseeker |
brlcad: OK, I see a failure |
| 13:07.22 |
starseeker |
looks like a different one though |
| 13:07.38 |
starseeker |
brlcad: are you doing an in-dir or out of dir
build? |
| 13:09.15 |
CIA-77 |
BRL-CAD: 03starseeker * r43552
10/brlcad/trunk/src/mged/chgview.c: Don't shadow basename |
| 13:11.16 |
starseeker |
god I wish we could switch to CMake |
| 13:12.26 |
starseeker |
out of dir autotools build can't find the
tclscripts to initialize the gui, from the looks of it |
| 13:13.49 |
starseeker |
bizarrely enough, if I start up with nu,
package require Itcl DOES succeed |
| 13:15.27 |
starseeker |
confound it, autopath STILL has /usr/lib at
the head of the auto_path search path |
| 13:21.18 |
dloman |
so, brlcad (autotools) doesn't like out of src
builds atm? |
| 13:22.00 |
starseeker |
oh, I builds OK but it doesn't seem to
run |
| 13:23.27 |
starseeker |
auto_path is getting all the paths set to the
build dir versions of tclscript paths, which doesn't work because
they're still in the src dir |
| 13:25.57 |
starseeker |
we can either copy the tclscripts over into
the build dir (which is a variation on what CMake does currently)
or tweak the path logic to point back to the src dir |
| 13:26.32 |
starseeker |
but that won't fix the issue of /usr/lib being
up front in the search path, which is an excellent way to pull in
system installed things and mix the tcl/tk environment up
royally |
| 13:30.10 |
CIA-77 |
BRL-CAD: 03starseeker * r43553
10/brlcad/branches/cmake/ (66 files in 21 dirs): MFC
r43552 |
| 13:33.10 |
starseeker |
nice little subtle issues, particularly when
we have to use a local tweaked version of something and there's a
vanilla system version getting pulled in instead |
| 13:40.23 |
starseeker |
saddles up |
| 13:56.42 |
``Erik |
<PROTECTED> |
| 14:14.45 |
dloman |
I love it. |
| 14:15.05 |
dloman |
the IT guys are trying to tell me that its not
possible to drive 4 monitors from two video cards. |
| 14:17.24 |
dloman |
stares at brlcad's 5 monitor
setup and laughs at IT. |
| 14:17.55 |
dli |
dloman, two video cards for 5
monitor? |
| 14:18.09 |
dloman |
brlcad actually has 3 cards |
| 14:18.39 |
dloman |
but the point remains the same. the IT guys
I'm emailing obviously dunt know what they are talking
about. |
| 14:18.41 |
dli |
dloman, so X-server can handle as many as you
can provide? |
| 14:19.07 |
dloman |
should |
| 14:22.00 |
dli |
dloman, nice to know |
| 14:22.51 |
dloman |
stumbled upon a USB to DVI converter (see link
above) that allows up to 6 external monitors :) |
| 14:22.57 |
dloman |
....i really want to try that out. |
| 14:24.07 |
dli |
dloman, I got a thinkpad with faulty video
card, maybe, I can drive up a monitor by USB, but I'm not
sure |
| 14:24.40 |
dloman |
Might be worth a look. |
| 14:24.51 |
dloman |
is the video card integrated or
dedicated? |
| 14:25.38 |
CIA-77 |
BRL-CAD: 03erikgreenwald * r43554
10/brlcad/trunk/src/libfb/if_ogl.c: event is a struct, not a
pointer |
| 14:29.08 |
dloman |
ahahaha, just found a MB that has 7 x16 PCIe
slots. Get 7 Eyefinity6's and thats... 42 monitors plus 6 on USB.
48 screens. Awesome. |
| 14:30.05 |
dloman |
Heh, i wonder if that would actually work
:) |
| 14:31.28 |
starseeker |
consideres the power demand
and winces slightly |
| 14:31.57 |
dloman |
If you can afford all those cards and
displays, the Powersupply becomes trivial :) |
| 14:32.09 |
``Erik |
mr fusion |
| 14:32.18 |
dli |
dloman, it's deciated video, but still
integrated on the mobo |
| 14:32.40 |
``Erik |
1.21 jiggawatts (whatever a jigga
is) |
| 14:32.54 |
dloman |
dli: suckage. You opened the case to ensure
the gfx card isn't removable? |
| 14:33.58 |
dloman |
``Erik: you're borderline racist, so be
careful :) |
| 14:34.11 |
``Erik |
O.o |
| 14:34.12 |
dli |
dloman, yes, I opened it several times :)
right now, the thinkpad works as a file server, and a wireless
router for my VoIP |
| 14:34.20 |
dloman |
ahaha, nice. |
| 14:35.12 |
dli |
dloman, the VoIP box supports only wired net,
so the thinkpad hooks VoIP to wifi. |
| 14:35.20 |
dloman |
well that usb2dvi thingy costs about 100
bucks, and you might be able to get an entire replacement laptop
(depending on how old it is) for that. |
| 14:35.42 |
dloman |
hates wifi. too
slow. |
| 14:36.36 |
CIA-77 |
BRL-CAD: 03starseeker * r43555
10/brlcad/branches/cmake/src/libfb/if_ogl.c: MFC r43554 |
| 14:37.27 |
dli |
dloman, then, forget it, it's a 4-year-old
core2duo level. |
| 14:37.53 |
dloman |
heh, i just replaced my old core2 |
| 14:38.09 |
dli |
dloman, wifi is orders faster than VoIP
requires |
| 14:38.37 |
dloman |
understood |
| 14:39.19 |
dloman |
and wifi is normally 'good enough' but the
wife and I work with large photoshop files and getting them on/off
the fileserver via wifi is.... annoying at best. |
| 14:40.28 |
dli |
dloman, does wifi-N help? |
| 14:40.37 |
dloman |
some. |
| 14:40.51 |
dloman |
we're in the process of wiring the house for
gig-e |
| 14:41.01 |
dloman |
so we can plug in when we need
speed. |
| 14:41.34 |
dli |
dloman, that's wonderful :( when I get the
budget to redo my home, I will see how to make giga-e possible at
home |
| 14:42.29 |
dloman |
we're doing it slowly. |
| 14:42.31 |
dli |
dloman, do you work with NURB
objects? |
| 14:42.36 |
dloman |
negative. |
| 14:42.49 |
dloman |
I know next to nothing about nurbs. |
| 14:42.57 |
dloman |
its on the long 'tolearn' list. |
| 14:44.32 |
dli |
dloman, brlcad suggests me to start with NURB
intersection problem. I'm too slow here to start :( |
| 14:47.56 |
CIA-77 |
BRL-CAD: 03erikgreenwald * r43556
10/brlcad/trunk/src/irprep/ir-X.c: size_t casting fixes |
| 14:51.46 |
CIA-77 |
BRL-CAD: 03erikgreenwald * r43557
10/brlcad/trunk/src/mged/clone.c: size_t casting |
| 14:52.13 |
CIA-77 |
BRL-CAD: 03erikgreenwald * r43558
10/brlcad/trunk/src/mged/dm-ogl.c: fill out entire struct in
table |
| 15:12.03 |
brlcad |
starseeker: cool, what platform gave you the
defined but unused failures? |
| 15:14.00 |
brlcad |
dloman: you'd saturate the bus before you
could drive that many, but it would be fun to see |
| 15:15.05 |
brlcad |
4 smaller displays are possible on one
dual-dual, or two big'uns |
| 15:15.28 |
brlcad |
dli: making any progress? |
| 15:15.48 |
starseeker |
brlcad: was on gentoo |
| 15:15.56 |
starseeker |
I'm not sure if it was pulling in system itcl
or not |
| 15:16.09 |
starseeker |
those friggin auto_path settings make it
problematic |
| 15:16.28 |
brlcad |
dli: atually the suggestion was based purely
on your interest and background -- otherwise, I would have
suggested something much smaller to start with :) |
| 15:16.38 |
brlcad |
starseeker: did it work once
installed? |
| 15:16.39 |
starseeker |
but the gui issue at least was clear - paths
were set to build dir, files were in src dir - boom |
| 15:17.02 |
starseeker |
didn't try an install - was working on regress
- but I would expect that it did |
| 15:17.25 |
brlcad |
k |
| 15:17.29 |
brlcad |
I'll run a test here too |
| 15:17.47 |
brlcad |
if install works, then I can at least still
tag release |
| 15:18.17 |
starseeker |
I'm trying to wait for the switch to cmake to
really mess with the auto_path settings - they should really
simplify down now that we're duplicating the install layout
everywhere |
| 15:18.42 |
brlcad |
true, but it also needs to be
relocatable |
| 15:18.54 |
starseeker |
sure |
| 15:19.05 |
brlcad |
so an "uninstalled" build tree should also
work -- that's the main concern |
| 15:19.06 |
starseeker |
the CMake build, in my testing, is already
relocatable |
| 15:19.42 |
brlcad |
hm |
| 15:20.05 |
starseeker |
I just meant we'd have one set of dir paths
that list all the tclscripts subdirs, and then use either build
root or install root to id them for auto_path |
| 15:21.13 |
starseeker |
plus, I've got to figure out how to strip
those /usr/lib based paths out of the front of the auto_path
list |
| 15:22.18 |
brlcad |
it's not build or install root, it'd be the
runtime root for it to be relocatable |
| 15:22.25 |
starseeker |
er, right |
| 15:22.40 |
starseeker |
the bu_brlcad_data/bu_brlcad_root
logic |
| 15:22.43 |
brlcad |
k |
| 15:23.28 |
brlcad |
shouldn't need to strip paths off auto_path --
someone is adding it, that's the point of attack |
| 15:23.58 |
starseeker |
right - I'm just concerned it might be tcl
itself initializing auto_path to those values |
| 15:24.04 |
starseeker |
at this point I don't know |
| 15:24.11 |
brlcad |
sounds to me like a system pkgIndex.tcl
getting loaded |
| 15:24.52 |
starseeker |
uh... it can't do that until it has paths to
load pkgIndex.tcl files from - that's why it needs that C init
process |
| 15:25.23 |
brlcad |
sure, but there are the built-ini *_Init()
routines in the libraries that are loaded |
| 15:25.37 |
starseeker |
nods |
| 15:25.39 |
brlcad |
if it pulls a system .so, then it conceivably
can modify the auto_path |
| 15:26.36 |
starseeker |
it would help if there was a debug setting to
let us see where package require was finding its .so
files |
| 15:26.46 |
starseeker |
(for a given package require
operation) |
| 15:29.02 |
brlcad |
there are ways to introspect tcl from the
interpreter |
| 15:29.15 |
brlcad |
"info loaded", "into library" |
| 15:29.31 |
brlcad |
man n info |
| 15:30.15 |
starseeker |
ah, excellent |
| 15:30.22 |
brlcad |
"package names" |
| 15:32.05 |
starseeker |
I have a hunch my gentoo box has a system
itcl/itk that happened to work, and it found that (since it didn't
have the auto_path information needed to bridge the divide between
build and src dirs, based on the gui behavior) |
| 15:32.39 |
starseeker |
would almost be better if we could somehow
reject libraries not in the "BRL-CAD expected" location for Tcl
packages |
| 15:32.58 |
starseeker |
would avoid the accidentally, silently working
issue |
| 15:54.58 |
CIA-77 |
BRL-CAD: 03erikgreenwald * r43559
10/brlcad/trunk/src/mged/ (edsol.c mged.h): de-const due to
possible deletion |
| 16:00.21 |
CIA-77 |
BRL-CAD: 03erikgreenwald * r43560
10/brlcad/trunk/src/mged/fbserv.c: Move the forward declaration
into the #if section that uses it, to avoid the "declared 'static'
but never defined" error. |
| 16:08.04 |
CIA-77 |
BRL-CAD: 03erikgreenwald * r43561
10/brlcad/trunk/src/proc-db/vegetation.c: size_t cast fix |
| 16:17.37 |
dloman |
``Erik: I think i found it. in
GSThread::start(), the pthread was being created in a detached
state. In ~GSThread, pthread_join() was called. According to the
man page, calling _join on a detached thread is a no no. |
| 16:19.09 |
CIA-77 |
BRL-CAD: 03davidloman * r43562
10/geomcore/trunk/src/utility/GSThread.cxx: Commented out the
pthread_join() call in GSThread::~GSThread. Man page says that a
pthread created in a detached state cannot be used as a sync point
via pthread_join() |
| 16:19.24 |
dloman |
see if that messes anything up on bsd and/or
mac |
| 16:29.39 |
*** join/#brlcad Stattrav
(~Stattrav@unaffiliated/stattrav) |
| 16:39.39 |
CIA-77 |
BRL-CAD: 03davidloman * r43563
10/geomcore/trunk/src/utility/GSUuid.cxx: Fix a bug where the
buffer used to export the uuid was being free'd prior to that
buffer being used to generate a std::string. |
| 17:18.06 |
dli |
brlcad, yes, I'm still interested on the NURB
intersection part. I will have more time now to work on it. Had to
go to hospital often in past weeks |
| 17:19.38 |
dli |
brlcad, at this stage, I'm still reading
through the ON wiki: http://wiki.mcneel.com/developer/opennurbs/home |
| 17:20.11 |
dli |
brlcad, still, I will report my success as
well as my failure to you when it's time |
| 17:21.18 |
brlcad |
naturally, I'd hope we can choose the best
development path so we achieve success instead of failur
;) |
| 17:22.02 |
brlcad |
that might mean not biting on one of the
hardest problems first, there are lots of places where some
progress could be made that would help get you familiarized with
the source code |
| 17:23.16 |
dli |
brlcad, sure, through evolution is good, but
the code base overall is huge, it might easier for me to start by
writing something new |
| 17:24.04 |
brlcad |
new vs old isn't as important as starting with
something very small |
| 17:24.25 |
brlcad |
nurbs intersection is not small ... |
| 17:24.38 |
dli |
brlcad, if you can point to some place, I
would be glad to see whether I can fix some small things in
parallel with my intersection problem |
| 17:24.48 |
brlcad |
so that was probably over-ambitious without
having touched any other code |
| 17:26.33 |
CIA-77 |
BRL-CAD: 03brlcad * r43564
10/brlcad/trunk/doc/README.Linux: include instructions to force
32-bit on platforms that default to 64-bit as well. |
| 17:26.47 |
brlcad |
hm, something came up just this week that is a
nice small task |
| 17:27.23 |
brlcad |
and it's in a similar section of library code
as the brep/nurbs primitive |
| 17:27.28 |
brlcad |
the revolve primitive |
| 17:28.57 |
dli |
brlcad, sounds good. is there a bug tracking
thread for this task? |
| 17:29.06 |
brlcad |
basically, it's a very new primitive so it
doesn't yet have support for matrix transformations (scaling,
translation, rotation) |
| 17:29.46 |
brlcad |
support for matrix transforms is in one
function, which revolve doesn't presently implement |
| 17:30.12 |
brlcad |
https://sourceforge.net/projects/brlcad/forums/forum/362510/topic/4372998 |
| 17:30.49 |
brlcad |
that's merely a day or two project, unlike
nurbs which is several weeks :) |
| 17:31.34 |
dli |
brlcad, so, rt_generic_xform(), that's
specific enough for me. I will report what I can do later |
| 17:31.53 |
brlcad |
rt_generic_xform() isn't sufficient, which was
my last comment |
| 17:32.13 |
brlcad |
have to implement rt_revolve_xform() similar
to rt_extrude_xform() |
| 17:32.46 |
brlcad |
src/librt/primitives/revolve/revolve.c and
src/librt/primitives/extrude/extrude.c respectively, with the
function listed in src/librt/primitives/table.c |
| 17:33.00 |
dli |
brlcad, I get some rough idea about what's
needed here, need to read the code first |
| 17:33.40 |
CIA-77 |
BRL-CAD: 03brlcad * r43565
10/brlcad/trunk/src/other/tkhtml/Makefile.am: |
| 17:33.40 |
CIA-77 |
BRL-CAD: the clean rule was overriding
autoconf's default clean rule, which calls |
| 17:33.40 |
CIA-77 |
BRL-CAD: clean-am. this should fix the build
where you follow a 32-bit build with a |
| 17:33.40 |
CIA-77 |
BRL-CAD: 64-bit built and vice versa. stale .o
build files were getting left in the |
| 17:33.40 |
CIA-77 |
BRL-CAD: .libs dir causing the build to
halt. |
| 17:33.53 |
brlcad |
working on that will introduce you to how
primitives are implemented, some basic structures, the callback
interface, and some light math |
| 17:34.03 |
brlcad |
all useful and relevant for working on
nurbs |
| 17:34.21 |
dli |
brlcad, let's see. :) |
| 17:38.12 |
brlcad |
starseeker: do you have a clean
build? |
| 17:38.24 |
brlcad |
``Erik: when was the last time you tried a
windows build? |
| 17:38.33 |
brlcad |
needs a windows build
test |
| 17:38.45 |
brlcad |
mac/linux are clean here |
| 17:40.24 |
brlcad |
and ``Erik, what can you tell me about the
libtie integration? working on the writeup |
| 17:48.54 |
``Erik |
tried one yesterday, broke a lot using msvc8,
will try another |
| 17:49.41 |
``Erik |
libtie's functions are in librt, bots that are
oriented or have normals should pass to tie if you set the
rt_min_tie (emulated rt_min_piece) |
| 17:50.13 |
*** join/#brlcad Stattrav
(~Stattrav@117.192.128.118) |
| 17:50.13 |
*** join/#brlcad Stattrav
(~Stattrav@unaffiliated/stattrav) |
| 17:50.15 |
``Erik |
the rt_min_tie is set to 4 billion and some
change right now pending more testing |
| 17:50.15 |
brlcad |
how does one set rt_min_tie? |
| 17:50.23 |
``Erik |
uh, -c I think? |
| 17:50.36 |
brlcad |
ah, so it'll auto-kick over to tie for 4M+
models now |
| 17:51.21 |
``Erik |
4m+ bot primitives |
| 17:51.28 |
brlcad |
right, okay |
| 17:51.32 |
``Erik |
0xffffffff |
| 17:51.39 |
brlcad |
ah, heh |
| 17:52.09 |
brlcad |
interesting, so no way to turn it off then or
is 0 special? :) |
| 17:52.19 |
``Erik |
from when that was an int, not a
size_t |
| 17:54.01 |
brlcad |
and rough performance impact on a 4M model is
what? 10%-500%? average 50%? |
| 17:55.06 |
``Erik |
didn't have a 4m primitive, at ~2000 it was
200% (100% gain), 200 was like 150% (+50%), 12 was 70%
(-30%) |
| 17:56.01 |
brlcad |
so on a real model, should see the time about
cut in half |
| 17:56.43 |
``Erik |
that was tesselating a sphere, the 'real'
models I tried were tesselations of lots of arbs, not NURBS
tesselations from an importer |
| 17:57.03 |
``Erik |
the benchmark numbers were from tesselated
spheres |
| 17:57.37 |
``Erik |
jabs cia |
| 17:58.17 |
``Erik |
0 disables tie as of 6 minutes ago |
| 17:59.01 |
brlcad |
hehe, awesome |
| 17:59.39 |
brlcad |
so I'll run a quick test on some bot model I
have, see what things look like |
| 18:00.07 |
``Erik |
there're a couple rough edges that need
cleaned up before I'm comfortable impacting customers with
it |
| 18:00.13 |
brlcad |
any other caveats or useful info other than
maybe not so hot for tiny models? |
| 18:00.36 |
``Erik |
on occasion, it misses :D |
| 18:00.47 |
brlcad |
rt_min_tie is coming up empty, what's the real
name? :) |
| 18:01.13 |
``Erik |
rt_bot_mintie |
| 18:01.23 |
brlcad |
thx |
| 18:01.24 |
``Erik |
(to mimic rt_bot_minpieces) |
| 18:05.36 |
CIA-77 |
BRL-CAD: 03brlcad * r43566
10/brlcad/trunk/TODO: |
| 18:05.36 |
CIA-77 |
BRL-CAD: attr command now sorts
alphabetically, but then other users reportedly |
| 18:05.36 |
CIA-77 |
BRL-CAD: also/still want sorting based on
creation order (because it makes it easy to |
| 18:05.36 |
CIA-77 |
BRL-CAD: diff and/or find new additions. need
a sorting option. kicking off an EDITOR |
| 18:05.36 |
CIA-77 |
BRL-CAD: should now be better too now that the
invocation has been re-reviewed recently. |
| 18:10.12 |
``Erik |
script.c:1414: warning: comparison between
signed and unsigned |
| 18:10.19 |
``Erik |
src/remrt/remrt.c:752: warning: comparison
between signed and unsigned |
| 18:11.10 |
CIA-77 |
BRL-CAD: 03brlcad * r43567
10/brlcad/trunk/TODO: WE ARE FREE OF COMPILATION WARNINGS!
woot. |
| 18:14.51 |
CIA-77 |
BRL-CAD: 03erikgreenwald * r43568
10/brlcad/trunk/src/librt/primitives/bot/bot.c: rt_bot_mintie=0 now
means do not use tie |
| 18:16.43 |
brlcad |
my 752 is an FD_MOVE line.. doesn't seem
right |
| 18:17.11 |
``Erik |
I know, I've been grepping around, fbsd might
have a bad header |
| 18:17.41 |
brlcad |
maybe line index from 0 and it's complaining
about the tcp_listen_fd int fd |
| 18:19.29 |
``Erik |
oohhhhh |
| 18:20.42 |
``Erik |
got it |
| 18:21.15 |
``Erik |
(#ifndef FD_MOVE ... we had our own for os's
like fbsd which don't provide) |
| 18:21.54 |
brlcad |
interesting |
| 18:24.38 |
brlcad |
looks like rt_bot_mintie is only exposed via
mged tcl var, not -c |
| 18:25.42 |
``Erik |
hm, I used rt_bot_minpieces as a template,
thought the librt tcl.c was called on rt's startup |
| 18:26.04 |
brlcad |
i added it |
| 18:26.34 |
brlcad |
it is a tcl var inside the tcl interp that
libtcl has running, but that's not exposed to rt |
| 18:26.53 |
brlcad |
they are manually wired to the
global |
| 18:27.07 |
``Erik |
ah |
| 18:27.56 |
``Erik |
win32 just built, had to remove regex.h from
raytrace.h (or update the include paths for most projects) and add
ws2_32.lib to a handful of converters |
| 18:29.55 |
brlcad |
yeah, regex.h inside raytrace.h doesn't sound
like a good idea |
| 18:30.14 |
brlcad |
should be an implementation detail, not public
api requirement |
| 18:31.30 |
brlcad |
poor CIA-77 |
| 18:31.39 |
starseeker |
``Erik: guess you're right, I'll have to do
the void thing |
| 18:32.56 |
CIA-77 |
BRL-CAD: 03erikgreenwald * r43569
10/brlcad/trunk/include/raytrace.h: conditionalize
regex.h |
| 18:35.04 |
starseeker |
brlcad: I'm still getting failures on Mac with
script.c from src/tab |
| 18:35.16 |
starseeker |
script.c:33:5: error: "__STDC_VERSION__" is
not defined |
| 18:35.26 |
starseeker |
script.c: In function
âyy_get_next_bufferâ: |
| 18:35.27 |
starseeker |
script.c:1389: warning: comparison between
signed and unsigned |
| 18:40.38 |
``Erik |
the windows comment in rt/do.c is due to
"initializer not static". might need to assign those in cm_set() as
a runtime dealie |
| 18:41.05 |
brlcad |
aha, that makes more sense |
| 18:41.16 |
brlcad |
those variables are in librt, so it can't bind
them |
| 18:41.22 |
CIA-77 |
BRL-CAD: 03erikgreenwald * r43570
10/brlcad/trunk/include/raytrace.h: remove regex.h for now, windows
would need most vcproj files updated. |
| 18:41.24 |
brlcad |
dll import suckage |
| 18:46.44 |
CIA-77 |
BRL-CAD: 03erikgreenwald * r43571
10/brlcad/trunk/src/remrt/remrt.c: fdset uses unsigned, so fix if
we need FD_MOVE defined |
| 18:52.33 |
CIA-77 |
BRL-CAD: 03brlcad * r43572
10/brlcad/trunk/src/rt/do.c: add support for -c
rt_bot_mintie |
| 18:53.22 |
CIA-77 |
BRL-CAD: 03erikgreenwald * r43573
10/brlcad/trunk/misc/win32-msvc8/ (5 files in 5 dirs): link
ws2_32.lib for ntohl/htonl |
| 19:00.18 |
CIA-77 |
BRL-CAD: 03brlcad * r43574
10/brlcad/trunk/src/nirt/ (command.c nirt.c nirt.h): add support
for setting rt_bot_mintie from within nirt, similar to
rt_bot_minpieces. add new -T option in addition to new bot_mintie
nirt command. |
| 19:08.24 |
CIA-77 |
BRL-CAD: 03brlcad * r43575
10/brlcad/trunk/NEWS: added -T and bot_mintie options to nirt that
correspond with controlling when erik's new support for optimized
BoT raytacing kicks on |
| 19:12.50 |
CIA-77 |
BRL-CAD: 03brlcad * r43576
10/brlcad/trunk/src/rt/do.c: expand the comment now that the cause
is known. need to set during runtime, not compiletime. |
| 19:13.28 |
CIA-77 |
BRL-CAD: 03brlcad * r43577
10/brlcad/trunk/misc/win32-msvc/CMakeLists.txt: back to not needing
libregex search dir |
| 19:25.46 |
CIA-77 |
BRL-CAD: 03brlcad * r43578
10/brlcad/trunk/NEWS: (log message trimmed) |
| 19:25.46 |
CIA-77 |
BRL-CAD: reword tie integration from
user-visible perspective. erik integrated the |
| 19:25.46 |
CIA-77 |
BRL-CAD: former 'libtie' high-performance
triangle intersection engine (tie) into librt |
| 19:25.46 |
CIA-77 |
BRL-CAD: as a way to get optimized BoT
raytracing. initial results showing a halfing |
| 19:25.46 |
CIA-77 |
BRL-CAD: reduction of raytrace time for models
at 2k+ triangles. erik added an |
| 19:25.47 |
CIA-77 |
BRL-CAD: 'rt_bot_mintie' option (exposed via
mged and rt -c) for toggling when the |
| 19:25.48 |
CIA-77 |
BRL-CAD: optimization kicks in. currently set
really high at 4M until further testing |
| 19:41.00 |
CIA-77 |
BRL-CAD: 03starseeker * r43579
10/brlcad/branches/cmake/src/ (libged/search.c
librt/search.c): |
| 19:41.01 |
CIA-77 |
BRL-CAD: Not in final form yet (for one thing,
the regex in db_plan_t will have to become |
| 19:41.01 |
CIA-77 |
BRL-CAD: void and be cast as needed) but this
can run searches now. Can't be committed |
| 19:41.01 |
CIA-77 |
BRL-CAD: to trunk until after release in this
form, but committing now in cmake to have |
| 19:41.01 |
CIA-77 |
BRL-CAD: it checked in somewhere |
| 19:45.23 |
CIA-77 |
BRL-CAD: 03starseeker * r43580
10/brlcad/branches/cmake/src/libged/search.c: Whoops, need memory
here too. |
| 19:52.04 |
dloman |
brlcad: do we have any Display port to DVI or
VGA adapters around the office? |
| 19:52.10 |
dloman |
(if yo uknow off hand) |
| 19:53.17 |
CIA-77 |
BRL-CAD: 03starseeker * r43581
10/brlcad/branches/cmake/ (include/raytrace.h src/librt/search.h):
Stick regex in the private search header for now... |
| 20:23.41 |
starseeker |
``Erik: bob says tcl is wonderful |
| 20:24.22 |
``Erik |
especially on windows? |
| 20:25.54 |
starseeker |
heh |
| 20:51.56 |
CIA-77 |
BRL-CAD: 03starseeker * r43582
10/brlcad/branches/cmake/ (4 files in 3 dirs): This approach keeps
the plan data structure out of raytrace.h, and thus isolates
regex. |
| 21:43.55 |
CIA-77 |
BRL-CAD: 03starseeker * r43583
10/brlcad/branches/cmake/ (23 files in 16 dirs): MFC
r43582 |
| 21:48.02 |
*** join/#brlcad ``Erik_
(Here@c-69-140-109-104.hsd1.md.comcast.net) |
| 21:58.38 |
CIA-77 |
BRL-CAD: 03starseeker * r43584
10/brlcad/trunk/ (7 files in 3 dirs): Put the new search routines
into trunk |
| 21:59.05 |
brlcad |
hope you cross-compile tested that
:) |
| 22:00.39 |
starseeker |
brlcad: just mac so far - working on
it |
| 22:00.55 |
starseeker |
had to re-implement the search . ...
stuff |
| 22:01.03 |
starseeker |
I think it behaves the way you wanted it to
now |
| 22:01.16 |
*** join/#brlcad Klebel
(~mk@169.229.55.243) |
| 22:02.11 |
brlcad |
I actually just hope it works with it being
injected right before the release, and isn't like that nirt "fix"
made right before release |
| 22:02.32 |
starseeker |
brlcad: sorry, I know it's a rotten
time... |
| 22:02.56 |
starseeker |
I can back it out in trunk and just work in
cmake branch |
| 22:03.49 |
CIA-77 |
BRL-CAD: 03starseeker * r43585
10/brlcad/branches/cmake/src/ (4 files in 2 dirs): MFC
r43584 |
| 22:03.53 |
brlcad |
committing is fine, you should just be extra
care and be testing more than usual |
| 22:04.24 |
starseeker |
nods - it's hot off the
press, I just now got it running, so I'm starting the testing
now |
| 22:05.03 |
starseeker |
if you've got a working compile, you might
check if the new behavior of (say) search . -maxdepth=0 does what
you expect |
| 22:05.34 |
starseeker |
I think we can squash that TODO item, but
since you spotted the issue confirmation would be good
:-) |
| 22:05.58 |
brlcad |
I can check it (and you should too given the
timing) |
| 22:06.04 |
starseeker |
oh, I am |
| 22:06.41 |
starseeker |
I just ment I wanted to make sure I had the
behavior you wanted, given how hard you had to work to explain it
to me :-P |
| 22:07.44 |
brlcad |
at release time, it becomes more like how
trunk development used to be -- trunk HEAD should be treated like
stable: changes tested on multiple platforms before commit, runtime
tested on everything |
| 22:07.49 |
brlcad |
I know |
| 22:08.12 |
brlcad |
I mean "you too" should be making extra sure
that all of search still works, maybe run through the documented
examples |
| 22:08.20 |
brlcad |
not just the new feature |
| 22:08.21 |
starseeker |
ah |
| 22:08.26 |
starseeker |
gotcha |
| 22:09.50 |
starseeker |
how |
| 22:09.59 |
starseeker |
bot.c isn't happy |
| 22:11.13 |
brlcad |
ah right, warnings |
| 22:11.22 |
brlcad |
compile had -w in effect |
| 22:11.53 |
brlcad |
perfect example :) |
| 22:13.34 |
starseeker |
got those, moving on... |
| 22:15.53 |
brlcad |
you mean you already got those? |
| 22:17.35 |
starseeker |
think so - just remove the unused and cast to
size_t for comparisons |
| 22:17.53 |
starseeker |
(don't want to mess with bot->faces...
types right now) |
| 22:18.26 |
brlcad |
have them fixed here too |
| 22:19.21 |
starseeker |
ah - feel free to stomp mine |
| 22:19.35 |
brlcad |
yours matched mine |
| 22:19.40 |
starseeker |
cool |
| 22:19.48 |
brlcad |
but you apparently weren't getting unused var
warnings like I have here |
| 22:20.00 |
starseeker |
really? you got more? |
| 22:20.05 |
brlcad |
yep |
| 22:20.09 |
starseeker |
weird |
| 22:20.25 |
brlcad |
to be expected |
| 22:21.32 |
brlcad |
one of the lessons from all the cleanup is
that even minor version number differences in gcc result in
different warnings, along with changes to compilation options,
32-bit vs 64-bit, optimized vs non-optimized |
| 22:22.24 |
brlcad |
and they're not a combination that is exactly
a superset, so there ends up being something like 3! possible
configurations |
| 22:22.35 |
brlcad |
3! or 4! |
| 22:25.04 |
starseeker |
nods |
| 22:30.06 |
_psilva |
:( |
| 22:30.21 |
starseeker |
_psilva: ? |
| 22:35.17 |
starseeker |
brlcad: my mac compile is still failing in
src/tab on script.c |
| 22:37.14 |
starseeker |
search completes all the examples from the man
page on potential |
| 22:38.23 |
starseeker |
brlcad: what version of flex and bison (or
lex/yacc) are you working with? |
| 22:55.22 |
starseeker |
oh, right |
| 22:55.34 |
starseeker |
can't really test well on the mac because of
that messed up install |
| 22:59.44 |
starseeker |
with autotools anyway... |
| 22:59.57 |
starseeker |
search looks ok with the cmake build on the
mac |
| 23:16.15 |
CIA-77 |
BRL-CAD: 03starseeker * r43586
10/brlcad/trunk/src/libged/search.c: Re-add support for the '.'
option (e.g. search . -name s*) but this time do it at the ged
level with post-processing of the full search. Also doesn't print
the leading '/' character for the '.' searches. |
| 23:17.27 |
CIA-77 |
BRL-CAD: 03brlcad * r43587
10/brlcad/trunk/src/conv/patch/patch-g.c: curious that thick_no
only shows up as a potential longjmp clobber var when compiling in
32-bit mode |
| 23:17.47 |
CIA-77 |
BRL-CAD: 03brlcad * r43588
10/brlcad/trunk/src/librt/primitives/bot/bot.c: damnits, found a
bug during release testing. disable the optimization so release can
proceed. |
| 23:17.51 |
CIA-77 |
BRL-CAD: 03brlcad * r43589
10/brlcad/trunk/TODO: resolve the crash post-release |
| 23:22.18 |
CIA-77 |
BRL-CAD: 03starseeker * r43590
10/brlcad/trunk/src/librt/primitives/bot/bot.c: Clear some warnings
on bot.c |
| 23:24.29 |
CIA-77 |
BRL-CAD: 03brlcad * r43591
10/brlcad/trunk/src/librt/primitives/bot/bot.c: quell unused var
warnings for the non-optimized case. split vars across
impls. |
| 23:26.07 |
_psilva |
gdc crunch sucks |
| 23:26.45 |
_psilva |
need more comp days from this |
| 23:32.00 |
*** join/#brlcad Klebel
(~mk@w72.RIC.Berkeley.EDU) |
| 23:53.03 |
brlcad |
sushi:~ morrison$ flex --version |
| 23:53.03 |
brlcad |
flex 2.5.35 |
| 23:53.03 |
brlcad |
sushi:~ morrison$ bison --version |
| 23:53.03 |
brlcad |
bison (GNU Bison) 2.3 |