Page 1 of 1

bf-blender / Linux (2005/02/07)

Posted: Mon Feb 07, 2005 1:58 pm
by gabio
Glad to get time to compile! cat herding and all, pfff.
comiled using scons, gcc 3.3.5 and glibc 2.3.4 on my new gentoo machine!
--------------------------
Relevant feature:
-bugfixes (2155, 2162, 2116, 2166, 2160, 2117, 2187, 2191, 2194, 2157
-some work on interface texture font support.
-bpython stuff:
---support for edge data in Blender.NMesh
---improved script registration and file and image selectors
---World.get/setMode were not registered, so could not be directly called
---Jean-Michel Soler updated his texture baker
---Campbell and Jean-Michel improved the bvh importer
-Joystick sensor in game engine.
-Added color filtering for raytrace transparency.
-Added a color picker and a 'Set VertCol' button to the Texture face panel
-Fixed old annoyance; enabling true Ortho render in Blender.
-NMesh: added face.sel and face.hide attributes to NMFaces

Commit:
-ton
Log:
Reading .blend from runtime closed file handle twice... seemed to crash in
windows? Weird, was a commit from april last year (Nathan) and nobody has
reported errors. (bug report 2155)

-sirdude
Log:
Added bFTGL to extern and updated the Makefiles. I'm guessing there will
need to be tweaks but it seems to work on my linux box. I haven't
touched any of the other build systems so those will need to be done.

We probably don't need all of this stuff but I figured better to add a little
too much then to little.

Kent

-ton
Log:
Bug 2162

Current cvs bug; the texture 'crop' lines were not drawn using correct
Panel/Window transform. Simple fix.

Log:
Bug fix 2116

Area lights with "Noise" dithering only worked for first area light, the
other ones then used wrong table... bug since area lights are in blender.

-sirdude
Log:
Added some extra comments about creating a new image format.

Kent

-phase
Log:
Modified FTGL for proper texturefont support.
Also fixed msvc6 (release only) projectfile to build and output correctly.

Log:
Modified interface texture font support, which should work nicely with the freshly added bFTGL library.
Also removed some redundant #include's on some files.

-sirdude
Log:
fixed animations using jpeg (bug # 2166)

Basically the deal was on the last fix (multiple opens/closes to a filehandle)
I changed the return values to match other image formats 0=fail and 1 = good
(was 1=fail 0=good before)
I Didn't update the animation code to see this so it was
thinking the first frame failed because it was looking for the old return code.

Kent

-ianwill
Log:
BPython:
- Stephane Soppera added long missed support for edge data in Blender.NMesh + related doc;
- Michael Reimpell improved script registration (fixes bug report #2160) and the file and image selectors in Blender.Window (improved with suggestions from Yann Vernier). They now suppport methods as callbacks;
- World.get/setMode were not registered, so could not be directly called (reported by Ken Hughes). Still needs some work to improve things, including docs.

Scripts:
- Jean-Michel Soler updated his texture baker based on input from Appolux;
- Campbell and Jean-Michel improved the bvh importer: faster, better float units scaling (by Campbell); supports Poser 3.01 files (by jms).

Thanks guys!

-kester
Log:
Don't fail on mesh objects with no faces.

Log:
Added Joystick sensor (from snailrose)

Log:
Don't delete the Python Controller's private dictionary between frames.

Log:
Make the KX_PolygonMaterial Python member variables writable.

-stiv
Log:
fix some undeclared parameters

-ton
Log:
Added color filtering for raytrace transparency.

http://www.blender3d.org/cms/Transparen ... 514.0.html

Choice was for using a single (new) filter value and have it working OK
with existing alpha.

-blendix
Log:
Added a color picker and a 'Set VertCol' button to the Texture face panel,
so the color of faces can be set in Uv Face Select Mode. Patch provided by
Carsten Wartmann (moved the buttons, and removed the rgb sliders).

-stiv
Log:
bugfix: #2117 Overflow in Armature.Bone.setName()

contributed by Joilnen B. Leite (pidhash).

-sirdude
Log:
Fix to get gcc 3.4.1 with scons working again.

-hos
Log:
Correcting the logic of the hotkey modifiers for select
non-manifold. If you want it so all three modifiers are needed
(e.g. ctrl-alt-shift-MKEY) use:

if ( !(~G.qual & (LR_SHIFTKEY | LR_ALTKEY | LR_CTRLKEY)) )

instead of

if ( G.qual & (LR_SHIFTKEY | LR_ALTKEY | LR_CTRLKEY) )

(which really allows any modifier, alone or in combination, to work).

-ton
Log:
Fixed old annoyance; enabling true Ortho render in Blender.

It used to be a simple hack, scaling lens with 100, and moving the camera
to the back with an equivalent amount.
Because of the hack, making it 100% compatible with older files I could not
achieve (yet?). To help reminding users, I've added a print when reading
old files with Ortho cameras.

Full description of how it works can be found here;
http://www.blender3d.com/cms/Render_changes.515.0.html

Log:
Bug fix (from own collection)

After using clip-safe pixeldraw from glutil.c (preview render, bg picture)
the current viewport had to be re-set. Not doing so caused a 1 pixel offset
for live updates buttons (on mouse over)

Log:
Bug fixes;

#2187: Append to active layer now uses the real active layer
#2191: Stars didnt show up correctly in envmap render
#2194: Startfame > Endframe in anim render crashed (now it gives error)

-aphex
Log:
- updated MSVC 6 projectfiles.
- removed 'using namespace std;' from SCA_Joystick.h
(gave compile error on VC6).

-hos
Log:
Doing the hotkey modifiers for select non-manifold the Ton way.

-ton
Log:
Bug found by Bassam; in certain occasions reading a file calls setscreen()
without the (ugly) global curarea being set.
Fixed the crash in his sample by nicely passing on 'current area' as
argument.

-ianwill
Log:
BPython:
- Fix related to bug #2157: crash in Blender.Image.image.reload() method. "G.sima" was not being checked for validity. Fix by Joilnen B. Leite (pidhash). Thanks!
http://projects.blender.org/tracker/?fu ... group_id=9
- Added optional argument to Blender.Object.object.getData() method: getData(only_name = True) or (only_name = 1) or (1) will return only the obdata name, not a wrapper for the given struct. This is a test based on a request by Campbell Barton who submitted code for his proposed .getDataName() method (thanks!).
- doc updates and small fixes.

-sirdude
Log:
Fixed typo Latice...

-ton
Log:
Fixed error in having Ztransp rendered in back of normal solid faces. Was
caused by changing definition of OSA pixel structs 3 weeks ago.
(result: AA looked bad).

Thanks phase for pointing at this.

-ianwill
Log:
BPython:
- NMesh: added face.sel and face.hide attributes to NMFaces, to set / get selection and visibility state of faces as they appear in edit mode.
- doc updates, including the right fix to two edge related methods, thanks to Stephane Soppera for pointing it (my fault, Stephane).

-ton
Log:
Parkinson bug! A & accidentally became &&...
Error caused exceptions for ortho render to be active always... making
some frontfaces and backfaces mixupped.

--------------------------
a new build is available

Posted: Tue Feb 08, 2005 1:43 pm
by mjordan
It seems depth buffer doesn't work correctly on Linux and Ati Mobility U1 (IGP320M) using X.org 6.8.1:
http://projects.blender.org/tracker/ind ... 9&atid=125

Posted: Tue Feb 08, 2005 3:52 pm
by Zsolt
Okay, I don't know where to post this, as its not exactly a "bug". It has taken away about three days of my time before I figured out what was wrong.

[img]zedman.uw.hu/images/precision_UIbug.jpg[/img]

This is in the Render panel, the Yafray GI tab, when the GI method is set to full and the Cache button is depressed. The problem is the tooltip of the "Prec" button, which tells you that the lower the number, the better and slower it is. NOT! I set it to the highest setting (50) for a large render, since I wanted it to finish as fast as possible. And it rendered and rendered and rendered. The rendertime was ~ 50 hours and it was about halfway finished! I set it back to the default value of 10, and it rendered in about 16 hours.
So please change the tooltip to read something like: "Precision, the higher the better and slower", as it can save a LOT of time.

Posted: Fri Feb 11, 2005 2:27 am
by Napoleon
Hi all, i run debian sid. the official blender 2.36 runs well on my system. I downloaded the testing build just to play with some of the new stuff. I extacted the tar.gz file which created a directory with all the files in it. I located the blender file and tried to run it but it just doesn't respond and i dont get any error messages either.

What could be the matter? Like i said, the official blender release runs well.

Posted: Fri Feb 11, 2005 2:32 pm
by paleajed
I don't know if this is the right place for this, but I suppose so.

First and for all, I am not using this test-build but built one from CVS myself at about the same date.

I am currently developing an edgeslide-plugin in BPython (http://www.blender.org/modules.php?op=m ... pic&t=5620). During development I noticed the class variable .flag of the new edge BPython object does not work correctly. When trying to test for the "SELECTED" flag I noticed the .flag variable was 0 for each and every edge. Still, several edges were selected. Maybe flag data is not implemented yet?[/url]

Posted: Sat Feb 12, 2005 1:38 am
by gabio
paleajed wrote:I don't know if this is the right place for this, but I suppose so.

First and for all, I am not using this test-build but built one from CVS myself at about the same date.

I am currently developing an edgeslide-plugin in BPython (http://www.blender.org/modules.php?op=m ... pic&t=5620). During development I noticed the class variable .flag of the new edge BPython object does not work correctly. When trying to test for the "SELECTED" flag I noticed the .flag variable was 0 for each and every edge. Still, several edges were selected. Maybe flag data is not implemented yet?[/url]
The best is to poste such related topic to the bf-python mailing list directly. Though here it's gonna work too.
Glad to see this feature around. I Think it could be good as a hardcoded feature. Sure this plugin gonna ease the integration later.
Continue the good work