Mon 05 Dec 2011 01:00:46 PM UTC, comment #16:
Yup, this is fixed. Thank you.
Am I right that all the new features I added in FB gui for handling of -j and -k and -X and -Y were removed ? I don't seem to be able to decide on geometry for the video window (which always seem to be re-rendering the whole available screen rather than just the movie portion)
|
Mon 05 Dec 2011 10:20:04 AM UTC, comment #15:
I'll test as soon as I'll have a chance. Meanwhile I've built on one system and got this warning which you may want to clean:
gui/fb/fb.cpp:495: warning: unused parameter ‘ranges’
|
Mon 05 Dec 2011 05:15:08 AM UTC, comment #14:
Double buffering with AGG in a framebuffer works again as of #3702f10. It's also enabled now by default since it does look better, but has a configure option to disable it.
|
Thu 29 Sep 2011 02:20:04 PM UTC, comment #13:
Depends when we do a release. If we do a release soon, then no. If we aim for a release in a few weeks, then maybe yes.
|
Thu 29 Sep 2011 10:12:47 AM UTC, comment #12:
I think this should be fixed before the release.
Do you agree, Rob ?
|
Tue 20 Sep 2011 01:17:52 PM UTC, comment #11:
I mean getting back what used to work: FB/AGG with double buffering.
I don't think any other GUI was using that configure switch (--enable-double-buffering).
GTK is likely using double buffering internally, no matter ENABLE_DOUBLE_BUFFERING define.
Double buffering for FB/OpenVG (if needed) can wait.
|
Tue 20 Sep 2011 01:11:55 PM UTC, comment #10:
You mean enabled only for AGG and the FB ?
|
Tue 20 Sep 2011 07:10:50 AM UTC, comment #9:
Can we have double-buffering only enabled for FB while we wait for the refactoring ? That way whoever uses FB doesn't suffer from the regression, and this bug can be converted into a enhancement request
|
Wed 07 Sep 2011 04:30:04 PM UTC, comment #8:
A regression is something that used to work and now doesn't.
|
Wed 07 Sep 2011 04:19:03 PM UTC, comment #7:
Like I said, double buffering now needs to be refactored, so I don't see this as a regression at all. It was never enabled by default, and none of our shipped packages nor releases ever had this enabled. Rather than being a bug it should be a future task.
|
Wed 07 Sep 2011 04:15:59 PM UTC, comment #6:
I've just checked. Double buffering in FB/AGG works as of Gnash 0.8.10dev (premerge-20729-71d22cb)
W/out double buffering the playback is horribly flickering, while with is a pleasure.
Being or not enabled by default doesn't change the fact that FB/AGG used to work nicely with double buffering and now isn't anymore. That's the regression part.
|
Wed 07 Sep 2011 04:07:46 PM UTC, comment #5:
It didn't work months ago when I tested it, but as all the code has changed heavily around it, I don't think it matters. This wouldn't be a regression anyway, as it was never enabled by default anyway. The double buffering code needs to be refactored so it works with hardware based buffering, which the current implementation failed to do.
|
Wed 07 Sep 2011 04:04:18 PM UTC, comment #4:
I'm not sure I can confirm double buffering not being implemented before. I belive I added that option after getting too much flickering.
Do you want me to check it or will you ?
|
Wed 07 Sep 2011 03:57:01 PM UTC, comment #3:
Yes, as it never worked anyway, and it'll be easy to add back in whenever somebody (maybe me) implements double buffering.
|
Wed 07 Sep 2011 03:55:06 PM UTC, comment #2:
I confirm I have ENABLE_DOUBLE_BUFFERING defined (as per --enable-doublebuf).
Should we take the chance to drop the configure switch as well ?
|
Wed 07 Sep 2011 03:46:21 PM UTC, comment #1:
The FB GUI built for me on 10 different GNU/Linux distributions, so I would check if you have ENABLE_DOUBLE_BUFFERING set. The original code for double buffering never was finished by whomever worked on that. To double buffer now will require some refactoring to support hardware accelerated graphics. That was left as a future task.
|
Wed 07 Sep 2011 03:26:29 PM UTC, original submission:
My usual configure line:
'--enable-maintainer-mode' '--prefix=/extra/gnash-dev' '--disable-jemalloc' '--enable-fps-debug' '--with-swfdec-testsuite=/usr/src/swfdec/swfdec/test' '--enable-http-testsuite=http://localhost/' '--enable-python' '--enable-extensions=fileio,mysql' '--enable-doublebuf' '--enable-red5-testing' '--disable-static' '--enable-gui=gtk,qt4,sdl,fb,dump' '--enable-media=ffmpeg,gst'
Triggers a build failure:
/usr/src/gnash/gnash-head/gui/fb/fb.cpp: In destructor ‘virtual gnash::gui::FBGu
i::~FBGui()’:
/usr/src/gnash/gnash-head/gui/fb/fb.cpp:182: error: ‘buffer’ was not declared in this scope
/usr/src/gnash/gnash-head/gui/fb/fb.cpp: In member function ‘virtual void gnash::gui::FBGui::renderBuffer()’:
/usr/src/gnash/gnash-head/gui/fb/fb.cpp:381: error: ‘var_screeninfo’ was not declared in this scope
/usr/src/gnash/gnash-head/gui/fb/fb.cpp:383: error: ‘_drawbounds’ was not declared in this scope
/usr/src/gnash/gnash-head/gui/fb/fb.cpp:396: error: ‘m_rowsize’ was not declared in this scope
/usr/src/gnash/gnash-head/gui/fb/fb.cpp:398: error: ‘fbmem’ was not declared in this scope
/usr/src/gnash/gnash-head/gui/fb/fb.cpp:398: error: ‘buffer’ was not declared in this scope
make[4]: *** [fb_gnash-fb.o] Error 1
I'm not sure which other architecture will have this problem, as the FB gui was disabled by default by a recent commit of Rob.
|