Wed 29 Nov 2006 12:05:16 PM UTC, comment #16:
it is possible that refactoring of the code fixed it as a side effect, yes.
Please keep trying in the future as since we don't have an automatic testcase for this it can as well break again :/
|
Wed 29 Nov 2006 11:37:42 AM UTC, comment #15:
Seems that I can't no more reproduce the bug of the missing background. Has something be modified in the source since then?
|
Mon 27 Nov 2006 12:47:09 PM UTC, comment #14:
Instead of uncovering you can resize the window, which should trigger a complete redraw of the scene!
|
Mon 27 Nov 2006 12:46:20 PM UTC, comment #13:
Hmm, the invalidated bounds are calculated when advance_movie() already succeeded.
Uncovering the window does not help in this case as it just copies the in-memory frame buffer to screen (no re-rendering).
It really seems that only the right girl has the m_invalidated flag set on startup or the other two girls are not even yet loaded.
m_invalidated is initialized with "true" in the "character" constructor and only display() does clear it (and in that case it should be visible).
The "Undress me" text below is an animation and thus is updated continously.
It's hard to analyze since I can't reproduce it.
Is it possible that the frame-loaded check is buggy?
|
Mon 27 Nov 2006 12:18:27 PM UTC, comment #12:
I can't anymore get the crash.
However I can again sometimes see the problem shown in file #11330.
When I have this problem I can't no more see the rest of the image, even when covering/uncovering the image with another window, as suggested in comment #7. However if I move the mouse cursor over the girls, the 3 girls appears.
|
Mon 27 Nov 2006 09:31:16 AM UTC, comment #11:
> do you know how Gnash syncs between rendering and parsing thread?
Sure, syncing is inside movie_instance::advance.
Basically, before advancing, it is waited until the
"next" frame is loaded (no matter if we are in STOP mode).
The wait is implemented with conditinal variables
and a mutex, which is locked/unlocked at each iteration
in the parser loop.
We haven't finished the discussion about invalidated
rects, btw...
|
Mon 27 Nov 2006 08:53:58 AM UTC, comment #10:
I doubt it's a problem with AGG. Probably it's related to the invalidated bounds computation and so is not related to the renderer.
Strk, do you know how Gnash syncs between rendering and parsing thread? I guess the problem is that the first frame is being rendered before it has been fully parsed. Or at least the invalidated bounds are calculated too early...
|
Mon 27 Nov 2006 08:45:45 AM UTC, comment #9:
mmm.. backtraces are not much useful if you don't build
with debugging symbols enabled.
Please rebuild with the -g switch
CXXFLAGS='-g -O0' make
(probably needs 'make clean').
This would include line numbers in the backtrace.
|
Mon 27 Nov 2006 08:34:31 AM UTC, comment #8:
Now I get the attached backtrace with this file.
(file #11367)
|
Thu 23 Nov 2006 02:12:13 PM UTC, comment #7:
Ok, seen strip.png now. It seems a problem with the "invalidated rect" computation, probably also related to bug #18186.
See if you can obtain a full redraw, maybe covering the window and exposing it again.
|
Thu 23 Nov 2006 01:37:52 PM UTC, comment #6:
strip.png is a png file. The other two are text files. And I am using a FC6.
|
Thu 23 Nov 2006 12:22:28 PM UTC, comment #5:
Savanna tells me your attachemnts are 'BIN' files,
are you on a mac ? Can you send 'jpegs' or 'png' ?
|
Thu 23 Nov 2006 10:00:26 AM UTC, comment #4:
Hey, it's appearing again! Appears only some time.
I attch a screenshot and output of gnash -v when works fine and not.
I am using CVS head on a FC6 with agg-2.5.
|
Wed 22 Nov 2006 02:11:59 PM UTC, comment #3:
I did a checkout but seems I get only new testcases.
I retried (from disk and network) but seems working fine now. Before I also experienced some strange text corruption on simpleguess, now also this is gone.
Ah, the original file was:
http://funcop.net/strip.swf
|
Wed 22 Nov 2006 12:04:41 PM UTC, comment #2:
What changed since you could reproduce it ?
If nothing it might be a timing problem, try again and again
and see if you can obtain different results.
Also try loading from local disk or from a remote one.
|
Wed 22 Nov 2006 11:26:44 AM UTC, comment #1:
Uhmm, I can't no more reproduce this now.
|
Wed 22 Nov 2006 10:21:48 AM UTC, original submission:
When rendering this file with Agg (current CVS) I can see only a girl and "Undress me!!!"; after moving the mouse over the others girl I can see them.
|