Wed 29 Aug 2007 11:13:03 AM UTC, comment #18:
Some unit testing for movie_root anyone ?
Like calling set_background_alpha() and check get_background_alpha() might be enough to reproduce
this problem on frnd() change ...
|
Wed 29 Aug 2007 07:48:50 AM UTC, comment #17:
Works for me.
Black background is due to a invisible background color (alpha=0 instead of alpha=255), see movie_root.h:
void set_background_alpha(float alpha)
{
m_background_color.m_a = iclamp(frnd(alpha * 255.0f), 0, 255);
}
|
Wed 29 Aug 2007 05:45:59 AM UTC, comment #16:
confirmed.
cvs-head works now, wasn't before strk's commit.
> Note that rint() is not the same as fchop().
It's funny that this little change caused such big regression(many of the swf files in our testsuite turned into black backgrounds).
|
Wed 29 Aug 2007 04:20:41 AM UTC, comment #15:
ahhh, indeed rint() seems to be handling things more arbitary.
|
Wed 29 Aug 2007 02:38:14 AM UTC, comment #14:
This change in particular introduced the regression:
- inline int frnd(float f) { return fchop(f + 0.5f); } // replace with inline asm if desired
+ inline int frnd(float f) { return (int)rint(f + 0.5f); } // replace with inline asm if desired
Note that rint() is not the same as fchop().
|
Wed 29 Aug 2007 02:15:01 AM UTC, comment #13:
2007-08-28 00:00 works, so the culprit is a change between 00:00 and 05:00. The only change in the ChangeLog between these two periods is an entry in Markus log:
> * libbase/math.h OS X has got xF().
Attached the patch which introduced the problem.
(file #13829)
|
Wed 29 Aug 2007 01:55:24 AM UTC, comment #12:
ChangeLog-2007-08-28-5:00 fails.
Current range is:
2007-08-28 05:00 <-- fails
2007-08-27 20:00 <-- works
|
Wed 29 Aug 2007 01:42:03 AM UTC, comment #11:
2007-08-28 15:00 fails.
|
Wed 29 Aug 2007 01:13:54 AM UTC, comment #10:
2007-08-27 20:00 works. Last ChangeLog entry:
2007-08-27 Deanna Phillips <deanna>
* testsuite/misc-ming.all/NetStream-SquareTestRunner.cpp: include
<unistd.h>
|
Wed 29 Aug 2007 12:59:50 AM UTC, comment #9:
2007-08-27-15:00 works. Last ChangeLog entry:
2007-08-27 Sandro Santilli <strk@keybit.net>
* server/cxform.cpp, server/font.cpp, server/matrix.cpp,
server/stream.h, server/parser/button_character_def.cpp,
server/parser/edit_text_character_def.cpp,
server/parser/filter_factory.cpp,
server/parser/sound_definition.cpp,
server/parser/video_stream_def.cpp,
server/swf/PlaceObject2Tag.cpp, server/swf/tag_loaders.cpp,
utilities/parser.cpp: read_uint(1) => read_bit();
* server/stream.cpp (read_uint): htf_sweet.swf aborts if we
assert bit_count <= 24, so change this base to <= 32 (will
inspect later).
|
Wed 29 Aug 2007 12:28:02 AM UTC, comment #8:
2007-08-27 works fine. gtk/agg. Last ChangeLog entry:
2007-08-26 Deanna Phillips <deanna>
* doc/gnash.1: Updated.
|
Tue 28 Aug 2007 05:39:36 PM UTC, comment #7:
By chance I noticed that the background color passed to begin_display() has the alpha (m_a) component set to zero. I'm pretty sure this is wrong...
|
Tue 28 Aug 2007 04:09:49 PM UTC, comment #6:
It is a recent regression. A SWF I often use to start gnash for testing suffers from the same problem, and it was fine at the weekend, I think until the 26th at least.
|
Tue 28 Aug 2007 02:05:48 PM UTC, comment #5:
Apparently any movie where you can see the default background (which is defined by a solid color in the SWF file) shows a black background. gravity.swf (in testsuite) suffers the same problem.
|
Tue 28 Aug 2007 01:13:45 PM UTC, comment #4:
simpleguess.swf has the same problem (swf4)
md5: 45c113fc0357358dc300b0041f22dfea
(should be somewhere in the tracker already)
|
Tue 28 Aug 2007 11:33:08 AM UTC, comment #3:
GTK-OGL with HEAD as of 2007/08/28 13:30 CEST: black background
|
Tue 28 Aug 2007 11:16:36 AM UTC, comment #2:
Yes, can reproduce:
GTK-AGG with HEAD as of 2007/08/28 11:58 CEST: black background
GTK-AGG with RELEASE 081 as of 2007/08/23 14:02 CEST: white background
GTK-OGL with HEAD as of 2007/08/23 12:13 CEST: white background
(still trying other variants)
|
Tue 28 Aug 2007 10:16:53 AM UTC, comment #1:
Can you reproduce the problem ?
|
Tue 28 Aug 2007 10:14:58 AM UTC, original submission:
http://foo.keybit.net/~strk/tmp/all.swf
The file above looks right with 0.8.1 (white background) and
wrong with head (black background)
|