mirrors/gstreamer
mirrors/gstreamer
1
1
Fork 0
mirror of https://gitlab.freedesktop.org/gstreamer/gstreamer.git synced 2024-12-24 09:10:36 +00:00
Code Issues 1 Projects Releases Wiki Activity
13690 commits 44 branches 1510 tags 533 MiB
Commit graph

3099 commits

Author SHA1 Message Date
Sebastian Dröge
c9b17f28f6 eglglessink: Make compilable on desktop Linux with MESA too 2012-10-18 14:35:52 +02:00
Sebastian Dröge
6fb5a9b716 eglglessink: Integrate into the build system 
Conflicts:
	ext/Makefile.am
 2012-10-18 14:35:49 +02:00
Sebastian Dröge
ebb99f27ac eglglessink: Fix debug output 2012-10-18 14:35:20 +02:00
Sebastian Dröge
61db9ad941 eglglessink: Also clear the surface once when the buffer content is preserved 2012-10-18 14:35:20 +02:00
Sebastian Dröge
147edd1fc2 eglglessink: Only draw black borders if the default swap behaviour of the surface is to destroy the buffer 2012-10-18 14:35:20 +02:00
Sebastian Dröge
a025c80b4c eglglessink: Remove unused label 2012-10-18 14:35:20 +02:00
Reynaldo H. Verdejo Pinochet
5bdf1ec099 eglglessink: Factor display region reset to its own func 2012-10-18 14:35:20 +02:00
Reynaldo H. Verdejo Pinochet
c3a12d737e eglglessink: Always Reset display_region.* at EGL/GLES ctx wipe 2012-10-18 14:35:20 +02:00
Reynaldo H. Verdejo Pinochet
ac037ce319 eglglessink: Factor ELG/GLES context cleanup into its own func 2012-10-18 14:35:20 +02:00
Reynaldo H. Verdejo Pinochet
ab81bd8f1e eglglessink: Drop drafted fast rendering path logic 2012-10-18 14:35:20 +02:00
Sebastian Dröge
9e8ad1097b eglglessink: Also update header file 2012-10-18 14:35:20 +02:00
Sebastian Dröge
f5043f5d64 eglglessink: Provide the texture positions to the shader on every frame too 2012-10-18 14:35:20 +02:00
Sebastian Dröge
9941466ab4 eglglessink: Draw black borders and also use a single GL buffer for all vertices 
Fixes weird artifacts outside the video frame on
Qualcomm Snapdragon chipsets and should be faster
than glClear() on the complete area all the time.
 2012-10-18 14:35:20 +02:00
Sebastian Dröge
a65ed4d7c1 eglglessink: Fix memory leak 2012-10-18 14:35:20 +02:00
Sebastian Dröge
f7a0da4da9 eglglessink: Remove useless debug output added for debugging earlier 2012-10-18 14:35:20 +02:00
Sebastian Dröge
0ac17bcbc2 eglglessink: Make window setting/handling a bit more threadsafe 2012-10-18 14:35:19 +02:00
Sebastian Dröge
e766bd1e0a eglglessink: Print window handle after setting it 2012-10-18 14:35:19 +02:00
Reynaldo H. Verdejo Pinochet
32861ab0a7 eglglessink: Move DAR query & store to surface init 
The DAR (display's pixel aspect ratio) should remain a
constant at runtime so there's no point on repeatedly
query and store this value. Doing it at surface setup
should be enough. As an added bonus this change should
make rendering a bit faster.
 2012-10-18 14:35:19 +02:00
Sebastian Dröge
406630a294 eglglessink: Don't signal the GCond too often 2012-10-18 14:35:19 +02:00
Sebastian Dröge
82dc2a3544 eglglessink: Add myself as author 2012-10-18 14:35:19 +02:00
Sebastian Dröge
87a028a0a4 eglglessink: Make sure to always reset the display region in start() 2012-10-18 14:35:19 +02:00
Sebastian Dröge
28fe3f7888 eglglessink: Also commit missing changes to the header file 2012-10-18 14:35:19 +02:00
Sebastian Dröge
610e33b049 eglglessink: Also include $(GST_BASE_CFLAGS) 2012-10-18 14:35:19 +02:00
Sebastian Dröge
caf9d1febd eglglessink: Fix race condition 2012-10-18 14:35:19 +02:00
Sebastian Dröge
99f738bbc0 eglglessink: Refactor rendering and all context specific GL things into a separate thread 
Also fixes some threading related problems.
 2012-10-18 14:35:19 +02:00
Sebastian Dröge
11ec917ee0 eglglessink: Re-add accidentially reverted commit 2012-10-18 14:35:19 +02:00
Sebastian Dröge
ba3274c8f0 eglglessink: Remove unneeded flow_lock 
Use the object lock in the few places necessary, everything
else is protected by other things already.

Also store the context directly inside the instance struct
instead of allocating some memory for it.
 2012-10-18 14:35:19 +02:00
Sebastian Dröge
2b3279c9b5 eglglessink: Fix debug output and constant names 2012-10-18 14:35:19 +02:00
Sebastian Dröge
1f0bf75075 eglglessink: start()/stop() is called in READY<->NULL in basesink 
But this should be called in PAUSED<->READY.
 2012-10-18 14:35:18 +02:00
Reynaldo H. Verdejo Pinochet
d2dcecfc8a eglglessink: Fix for outbound DAR reported by EGL 
Some EGL implementations don't honor the spec
requirement of returning DAR values as w/h *
EGL_DISPLAY_SCALING.

This changeset:

- Fixes rendering on the Samsung Galaxy III
- Fixes wrong check on required 1.2 EGL version
- Reorders the code a bit.
 2012-10-18 14:35:18 +02:00
Sebastian Dröge
dde62cee8e eglglessink: Fix crashes if there are no EGL or GLES extensions 2012-10-18 14:35:18 +02:00
Reynaldo H. Verdejo Pinochet
a317f6936c eglglessink: Fix crash at element instantiation 2012-10-18 14:35:18 +02:00
Reynaldo H. Verdejo Pinochet
af009b7de7 eglglessink: GLSL: Save one extra texture lookup 2012-10-18 14:35:18 +02:00
Reynaldo H. Verdejo Pinochet
0ed02803f7 eglglessink: GLSL: Minor named index access tweak xyz -> rgb 2012-10-18 14:35:18 +02:00
Reynaldo H. Verdejo Pinochet
5c2b1dc631 eglglessink: GLSL: Assorted fixes to optimized code 
- Don't copy over opos as it's already a vec2
- Fix index order access at frag_YUY2_YVYU_UYVY_prog
 2012-10-18 14:35:18 +02:00
Reynaldo H. Verdejo Pinochet
73d865091c eglglessink: GLSL: Planar YUV converters optimization 
- Use consts for the transform's offset and cofficients values
- Use dot product instead of mult and add
- Avoid unneeded texture lookups
 2012-10-18 14:35:18 +02:00
Reynaldo H. Verdejo Pinochet
846e77871b Geglglessink: LSL: Optimize frag_YUY2_YVYU_UYVY_prog 
- Avoid repeatedly performing the texture lookup
- Use consts for the transform's offset and cofficients values
- Use the dot product instead of the explicit mult and add
 2012-10-18 14:35:18 +02:00
Reynaldo H. Verdejo Pinochet
eb8f7d63d5 eglglessink: GLSL: Drop unneded nxy vec2 2012-10-18 14:35:18 +02:00
Reynaldo H. Verdejo Pinochet
bb0b872644 eglglessink: Fix missing double quotation mark in GLSL code 2012-10-18 14:35:18 +02:00
Reynaldo H. Verdejo Pinochet
a95834a491 eglglessink: GLSL: Optimize frag_AYUV_prog 
- Avoid repeatedly performing the texture lookup
- Use consts for the transform's offset and cofficients values
- Use the dot product instead of the explicit mult and add
 2012-10-18 14:35:18 +02:00
Reynaldo H. Verdejo Pinochet
a7a2dcc433 eglglessink: Documentation: GstEglGlesSink struct 2012-10-18 14:35:18 +02:00
Reynaldo H. Verdejo Pinochet
38ca631a7a eglglessink: Documentation: GstEglGlesBuffer struct (DRAFT) 2012-10-18 14:35:17 +02:00
Reynaldo H. Verdejo Pinochet
c27c11582e eglglessink: Documentation: GstEglGlesImageFmt struct 2012-10-18 14:35:17 +02:00
Reynaldo H. Verdejo Pinochet
945b11fd02 eglglessink: Documentation: GstEglGlesRenderContext struct 2012-10-18 14:35:17 +02:00
Reynaldo H. Verdejo Pinochet
d97b3dba60 eglglessink: Assorted doc and reordering fixes 2012-10-18 14:35:17 +02:00
Reynaldo H. Verdejo Pinochet
031b6a57ac eglglessink: Documentation: Element properties 
Brief explanatory comments plus some reordering
to group default setting of properties and runtime
flags values.
 2012-10-18 14:35:17 +02:00
Sebastian Dröge
81463a1cb7 eglglessink: Add support for YVYU 2012-10-18 14:35:17 +02:00
Reynaldo H. Verdejo Pinochet
95a223a030 eglglessink: Documentation: GLSL Shaders 
Brief explanatory comments plus some rerdering
to group packed/planar converters.
 2012-10-18 14:35:17 +02:00
Reynaldo H. Verdejo Pinochet
731be48dac eglglessink: Documentation: Element overview 2012-10-18 14:35:17 +02:00
Reynaldo H. Verdejo Pinochet
1eef329725 eglglessink: Refactor REORDER shader. 
Avoid unneededly calling texture2D() 3 consecutive times.
 2012-10-18 14:35:17 +02:00
Sebastian Dröge
c91238a2af eglglessink: Fix handling of caps changes by only releasing what needs to be released 
And not terminating the display connection
 2012-10-18 14:35:17 +02:00
Reynaldo H. Verdejo Pinochet
310e5111b7 eglglessink: Move display_region reset to locked section 2012-10-18 14:35:17 +02:00
Reynaldo H. Verdejo Pinochet
884e5fa173 eglglessink: Reset display_region dims at caps renego 2012-10-18 14:35:17 +02:00
Reynaldo H. Verdejo Pinochet
f28f5df29e eglglessink: Cosmetics 2012-10-18 14:35:17 +02:00
Reynaldo H. Verdejo Pinochet
c255aa5dd4 eglglessink: Don't access local value from struct 2012-10-18 14:35:16 +02:00
Reynaldo H. Verdejo Pinochet
2ab9de8e4f eglglessink: Terminate display connection at caps renego 2012-10-18 14:35:16 +02:00
Reynaldo H. Verdejo Pinochet
734ac85bb4 eglglessink: Cosmetics 2012-10-18 14:35:16 +02:00
Sebastian Dröge
e022fc5389 eglglessink: Only attach the context once to the streaming thread 2012-10-18 14:35:16 +02:00
Sebastian Dröge
09b3518bae eglglessink: Make the context current for the thread and unbind it again around every use 
This fixes some crashes and leaking of our context into other
OpenGL code on the same thread.
 2012-10-18 14:35:16 +02:00
Sebastian Dröge
113253b292 eglglessink: Improve cleanup 2012-10-18 14:35:16 +02:00
Sebastian Dröge
70d077acd8 eglglessink: Remove unused label 2012-10-18 14:35:16 +02:00
Sebastian Dröge
f058b5d78b eglglessink: Allow setting a NULL handle 2012-10-18 14:35:16 +02:00
Sebastian Dröge
fc0dc0055a eglglessink: Revert "Do not update_surface_dims() once per frame" 
This reverts commit 82103376ede2bc7c63dd699f5982168660f67852.

We have to update the surface dimensions on every frame to
make sure that we get notified about updates. This happens
for example if the Android window gets re-layouted because
a widget takes more space than before.
 2012-10-18 14:35:16 +02:00
Reynaldo H. Verdejo Pinochet
a818d5c771 eglglessink: Do not update_surface_dims() once per frame 
And.. Upgrade debug message to warging on weird DAR.
Correct fix to avoid spaming with this message
once per each frame is avoid calling update_surface_dims()
at render_and_display(). This is handled at egl surface
init.
 2012-10-18 14:35:16 +02:00
Sebastian Dröge
162f63e064 eglglessink: And downgrade another warning too 2012-10-18 14:35:16 +02:00
Sebastian Dröge
7307ddd54b eglglessink: Downgrade GST_WARNING to GST_DEBUG 2012-10-18 14:35:16 +02:00
Sebastian Dröge
8dabfbfca8 eglglessink: Add some {} to if statements make them work as expected 2012-10-18 14:35:16 +02:00
Reynaldo H. Verdejo Pinochet
20a05726b3 eglglessink: Drop stale defines for max frame size 2012-10-18 14:35:16 +02:00
Reynaldo H. Verdejo Pinochet
a3530d9a3d eglglessink: Small commentary fixes here an there 2012-10-18 14:35:15 +02:00
Reynaldo H. Verdejo Pinochet
2b706d3fbe eglglessink: Avoid setting height or width twice 2012-10-18 14:35:15 +02:00
Reynaldo H. Verdejo Pinochet
eeb3b84d07 eglglessink: Use PAR/DAR to set viewport 2012-10-18 14:35:15 +02:00
Sebastian Dröge
c3fb44aef3 eglglessink: Correctly resize if the surface dimensions change 2012-10-18 14:35:15 +02:00
Sebastian Dröge
3fcea0dc5e eglglessink: Update surface dimensions on every buffer 2012-10-18 14:35:15 +02:00
Sebastian Dröge
2b2ae0fd6f eglglessink: More consistent naming of shader related variables 2012-10-18 14:35:15 +02:00
Sebastian Dröge
1ad253b9b7 eglglessink: Some variable name cleanup 2012-10-18 14:35:15 +02:00
Sebastian Dröge
e8f4829c13 eglglessink: Remove default-width/height properties 
These are completely unneeded
 2012-10-18 14:35:15 +02:00
Sebastian Dröge
985a185ffc eglglessink: Fix double free() 2012-10-18 14:35:15 +02:00
Sebastian Dröge
b5b98fc874 eglglessink: Don't leak EGL/ES render context 2012-10-18 14:35:15 +02:00
Reynaldo H. Verdejo Pinochet
bab81de275 eglglessink: Put EGL/GLES context vars on their own struct 2012-10-18 14:35:15 +02:00
Reynaldo H. Verdejo Pinochet
5526737ec8 eglglessink: Small TODO reformatting 2012-10-18 14:35:15 +02:00
Reynaldo H. Verdejo Pinochet
e8ed9e1097 eglglessink: Add TODO file 2012-10-18 14:35:15 +02:00
Sebastian Dröge
2323c40f1b eglglessink: Correctly implement ::expose() 
We don't need the past buffer for this, the old one is still available
as texture and we just need to draw our rectangle again.
 2012-10-18 14:35:15 +02:00
Sebastian Dröge
18e473307f eglglessink: Allow setting another window for the next time we reconfigure everything 2012-10-18 14:35:14 +02:00
Sebastian Dröge
aa463f0274 eglglessink: Also unref current caps after usage 2012-10-18 14:35:14 +02:00
Sebastian Dröge
c2de59b19c eglglessink: Implement support for caps changes 2012-10-18 14:35:14 +02:00
Sebastian Dröge
d91621d2ce eglglessink: More cleanup and making the element reusable 2012-10-18 14:35:14 +02:00
Sebastian Dröge
56b711bd29 eglglessink: Properly init/release resources when going from NULL->READY and the other way around 2012-10-18 14:35:14 +02:00
Sebastian Dröge
15c93da54e eglglessink: Generate the sink caps exactly once 2012-10-18 14:35:14 +02:00
Sebastian Dröge
ad6e314a6e eglglessink: Fix code style a bit to stop confusing indent 2012-10-18 14:35:14 +02:00
Sebastian Dröge
4e637ddb7b eglglessink: Fix support for YUY2 and UYVY 2012-10-18 14:35:14 +02:00
Sebastian Dröge
9e094541b0 eglglessink: Some property cleanup 2012-10-18 14:35:14 +02:00
Sebastian Dröge
25eb488f30 eglglessink: Add support for the remaining RGB formats 2012-10-18 14:35:14 +02:00
Sebastian Dröge
11368b49b2 eglglessink: Add support for Y444, Y42B and Y41B, just because it's so easy 2012-10-18 14:35:14 +02:00
Sebastian Dröge
d98d5c657e eglglessink: Add some simple TODOs 2012-10-18 14:35:14 +02:00
Reynaldo H. Verdejo Pinochet
98b9862167 eglglessink: Lock while altering display region coords 2012-10-18 14:35:14 +02:00
Reynaldo H. Verdejo Pinochet
a2fce44139 eglglessink: Fix _get_compat_format_from_caps() logic 
It's a getter so it shouldn't messing with the sink context.
Adding insult to injury it was setting selected_fmt without
locking.
The associated changes to the buffer functions are merely
make-it build fixes. Those functions are barely drafted
and in WIP state.
 2012-10-18 14:35:13 +02:00
Reynaldo H. Verdejo Pinochet
47d0505f40 eglglessink: Drop unneded initial set to NULL 2012-10-18 14:35:13 +02:00
Reynaldo H. Verdejo Pinochet
0c681dc2a6 eglglessink: Minor formatting and comment fixes 2012-10-18 14:35:13 +02:00
Reynaldo H. Verdejo Pinochet
63d8cd88c7 eglglessink: Clean-up gl/egl allocations at _stop() 
Also, conditionaly destroy window if the sink created its own.
 2012-10-18 14:35:13 +02:00
Sebastian Dröge
99765c3333 eglglessink: Add support for NV12 and NV21 
Also YUY2 and UYVY but this is disabled for now as it
doesn't work completely yet. Something is wrong with
the chroma.
 2012-10-18 14:35:13 +02:00
Sebastian Dröge
72d05a681f eglglessink: Add support for I420 and YV12 color formats 2012-10-18 14:35:13 +02:00
Sebastian Dröge
d3995c1a7b eglglessink: Preparations for supporting multiple textures for more YUV formats 2012-10-18 14:35:13 +02:00
Sebastian Dröge
3f734aafc7 eglglessink: Use the attrib locations as really used inside the shader 2012-10-18 14:35:13 +02:00
Sebastian Dröge
24261a4fe7 eglglessink: Really decouple vertex and texture positions in the shader 2012-10-18 14:35:13 +02:00
Sebastian Dröge
dafcdb0da3 eglglessink: Decouple texture and vertex positions 2012-10-18 14:35:13 +02:00
Sebastian Dröge
cb034b2ecd eglglessink: Reorder vertex value order to make a bit more sense 2012-10-18 14:35:13 +02:00
Reynaldo H. Verdejo Pinochet
8018681187 eglglessink: Handle frame scaling to fill surface 2012-10-18 14:35:13 +02:00
Reynaldo H. Verdejo Pinochet
90d5b76dc5 eglglessink: Draft set_render_rectangle() from the xOverlay interface 2012-10-18 14:35:13 +02:00
Reynaldo H. Verdejo Pinochet
f8b69a7c7f eglglessink: Remove stale comment 2012-10-18 14:35:13 +02:00
Reynaldo H. Verdejo Pinochet
1c008e95b3 eglglessink: Move VBO setup to setcaps 2012-10-18 14:35:12 +02:00
Reynaldo H. Verdejo Pinochet
d56d126031 eglglessink: Don't do context ops at render 
We don't need to repeatedly call
eglMakeCurrent() while using the
same context anyway.
 2012-10-18 14:35:12 +02:00
Reynaldo H. Verdejo Pinochet
ca74f676fd eglglessink: Don't clear surface color at render 2012-10-18 14:35:12 +02:00
Sebastian Dröge
9e23d0fc87 eglglessink: Rename shader variables to make more sense 2012-10-18 14:35:12 +02:00
Sebastian Dröge
d42d5e1139 eglglessink: Add support for AYUV 2012-10-18 14:35:12 +02:00
Sebastian Dröge
c4643491c4 eglglessink: Reorder caps a bit to make more sense 2012-10-18 14:35:12 +02:00
Sebastian Dröge
6adf608c99 eglglessink: Set rank to PRIMARY 2012-10-18 14:35:12 +02:00
Sebastian Dröge
1c58d921ff eglglessink: Add support for RGBA 2012-10-18 14:35:12 +02:00
Sebastian Dröge
930b4c179e eglglessink: Correctly generated caps for the supported formats 2012-10-18 14:35:12 +02:00
Sebastian Dröge
f9a5fc255d eglglessink: Add ::get_caps() vfunc that returns the actually supported caps 2012-10-18 14:35:12 +02:00
Sebastian Dröge
86ebf86433 eglglessink: Also require a configuration with EGL_SURFACE_TYPE containing EGL_WINDOW_BIT 2012-10-18 14:35:12 +02:00
Sebastian Dröge
7e8cc569dd eglglessink: Fix compiler warnings 2012-10-18 14:35:12 +02:00
Reynaldo H. Verdejo Pinochet
19c3b4d892 eglglessink: Clear surface to black by default 2012-10-18 14:35:12 +02:00
Reynaldo H. Verdejo Pinochet
19739bfb74 eglglessink: Make sure we get a config we can use with GLES2 2012-10-18 14:35:12 +02:00
Reynaldo H. Verdejo Pinochet
24b926a6cc eglglessink: set_window_handle() has been tested to work now 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
7ac284b1a1 eglglessink: Avoid autoscaling by default 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
a81b036355 eglglessink: Avoid repeatedly setting 2D resizing params 
Moved to initial texture setup so it hapens
just once while on the same context.
 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
a96fa541ee eglglessink: Avoid initializing EGL at _set_window_handle() 
_start() should take care of that.
This change asumes _setcaps() will
happen always after _start() is
done.
 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
ead20f4f0b eglglessink: Don't ask for a window if we already have one 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
22eb374b57 eglglessink: Fix EGL attribs var naming 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
c2cf1fa639 eglglessink: Fix slow path rendering on Android 
Specify missing and mandatory params for npot
Before calling glTexImage2D.

In particular, _WRAP_T and _WRAP_S need to be
set to GL_CLAMP_TO_EDGE to avoid all-black
rendering with npot sized frames.
 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
f33167f95b eglglessink: Add test on usable eglconfigs returned 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
303a346fc2 eglglessink: Drop stale error_locked label 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
5763c8d308 eglglessink: Cleanup stale code comments 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
391137a20e eglglessink: GlViewPort() should use actual surface size 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
f707961183 eglglessink: Avoid reiniting EGL after the first time 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
75fe18200a eglglessink: Use correct template caps 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
e439c01382 eglglessink: Fix missing texture setup bits 2012-10-18 14:35:11 +02:00
Reynaldo H. Verdejo Pinochet
68eac71910 eglglessink: Add error checking on eglSwapBuffers 2012-10-18 14:35:10 +02:00
Reynaldo H. Verdejo Pinochet
f8dcc8aaa5 eglglessink: Fix stale&wrong 0.5 alpha blend 2012-10-18 14:35:10 +02:00
Reynaldo H. Verdejo Pinochet
12d66a7d54 eglglessink: Add float precision specifier to fragment shader 
This is required by GLES and fixes compilation of fragment
shader prog.
 2012-10-18 14:35:10 +02:00
Reynaldo H. Verdejo Pinochet
28aadcad4e eglglessink: Add log reporting at shader compilation failure 2012-10-18 14:35:10 +02:00
Reynaldo H. Verdejo Pinochet
b4887425cc eglglessink: Add some missing checks on shader ops 
- Check whether there is shader compiler support
  and bail out otherwise.
- Add checks on compile and linking ops
 2012-10-18 14:35:10 +02:00
Reynaldo H. Verdejo Pinochet
2a84a72fe7 eglglessink: Fix setup chain logic 
Mostly a procedural reordering. Idea is to avoid
trying to setup a surface without a display.This
change relies on the asumption that:

1.- set_window_handle() can happen at any time
2.- setcaps will never get called before _start()
    is done.

If both conditions are meet the new procedure
should work as expected.
 2012-10-18 14:35:10 +02:00
Reynaldo H. Verdejo Pinochet
40d6cc2099 eglglessink: WIP: Reorder locking 2012-10-18 14:35:10 +02:00
Reynaldo H. Verdejo Pinochet
289f388085 eglglessink: Fix xOverlay logic 
Several fixes dealing with correct window
creation and set_handle() ops. Tested on
X11/mesa with internal window creation.
 2012-10-18 14:35:10 +02:00
Reynaldo H. Verdejo Pinochet
9e18b29be2 eglglessink: Actually test EGL configs before claiming support 2012-10-18 14:35:10 +02:00
Reynaldo H. Verdejo Pinochet
59b1870ac1 eglglessink: Caps nego EGL fixes 
- Fix alloc lol on format list: Asked for 2 slices and used 3.
- Reordered display & context init logic to prepare for actual
  config checking before claiming support for a given format.
  Change deals basically with setting up EGL display at start()
  and choose an EGL config only at setcaps time.
 2012-10-18 14:35:10 +02:00
Reynaldo H. Verdejo Pinochet
a8f60690aa eglglessink: Caps nego. Start extending RGB support to a set. 2012-10-18 14:35:10 +02:00
Sebastian Dröge
bd28c42fae eglglessink: Fix build of static plugin 2012-10-18 14:35:10 +02:00
Reynaldo H. Verdejo Pinochet
a83dae1b54 eglglessink: Drop wrong format test on _setcaps (draft leftover) 2012-10-18 14:35:10 +02:00
Reynaldo H. Verdejo Pinochet
75d4bf420a eglglessink: Support RGB24 by default 2012-10-18 14:35:10 +02:00
Reynaldo H. Verdejo Pinochet
1d77bf6c0f eglglessink: Instrument EGL error reporting 2012-10-18 14:35:09 +02:00
Reynaldo H. Verdejo Pinochet
b8c83f98b7 eglglessink: Correctly return GstFlowReturn from render_and_display() 2012-10-18 14:35:09 +02:00
Reynaldo H. Verdejo Pinochet
cca08a02a1 eglglessink: Move VBO setup to it's own function 2012-10-18 14:35:09 +02:00
Reynaldo H. Verdejo Pinochet
3fce24549f eglglessink: Fix render_and_display logic 
Switch procedure according to enabled rendering path
 2012-10-18 14:35:09 +02:00
Reynaldo H. Verdejo Pinochet
ada5a83eb4 eglglessink: Fix sample launch line with disabled window creation 2012-10-18 14:35:09 +02:00
Reynaldo H. Verdejo Pinochet
086563195c eglglessink: Rename stale android leftover 2012-10-18 14:35:09 +02:00
Reynaldo H. Verdejo Pinochet
ea03067f3f eglglessink: Move eglgles Sink to ext/ #2 2012-10-18 14:35:09 +02:00
Reynaldo H. Verdejo Pinochet
85250eb9e2 eglglessink: Move eglgles Sink to ext/ 2012-10-18 14:35:09 +02:00
Tim-Philipp Müller
32ba17cd0f Use gst_element_class_set_static_metadata() 
where possible. Avoids some string copies. Also re-indent
some stuff. Also some indent fixes here and there.
 2012-10-17 17:46:34 +01:00
Sergey N. Gorshkov
6d0d209a28 rtmpsink: handle RTMP_Write() return value correctly 
Error might also be negative (-1). Unclear if 0 should
be fatal as well though.

https://bugzilla.gnome.org/show_bug.cgi?id=681111
https://bugzilla.gnome.org/show_bug.cgi?id=686009
 2012-10-12 23:29:53 +01:00
David Régade
65add5533a rtmpsink: fix memory leak from URI verification via RTMP_ParseURL() 
In gst_rtmp_sink_uri_set_uri(), a test is performed in order
to be sure uri is correct for librtmp. This test calls
RTMP_ParseURL with 3 AVal pointers as parameters: host,
playpath and app.

AVal is a struct with a char* + int. After RTMP_ParseURL call,
host.av_val and app.av_val both refer a substring of "uri". But
playpath.av_val may be the result of a malloc so it needs to
be freed.

https://bugzilla.gnome.org/show_bug.cgi?id=681459
 2012-10-12 23:09:06 +01:00
Mark Nauwelaerts
304cf8e1c5 dtsdec: fix buffer parsing 
... to avoid feeding empty buffers to decoder.
 2012-10-12 16:48:54 +02:00
Wim Taymans
380b23af90 replace some playbin2 -> playbin 2012-10-10 14:14:43 +02:00
Michael Smith
0c2159a220 meta info: register thread-safely. 2012-10-03 11:05:06 -07:00
Jan Schmidt
6e4a9e2e24 resindvd: Remove caps lock from parsetter element. 
All the members it used to protect in 0.10 are protected by
the stream lock now.
 2012-10-02 22:20:20 +10:00
Jan Schmidt
6122e07795 resindvd: Update overridden caps when the DAR changes 
Send a new caps event downstream when we receive new orders
from the VM, if there are existing caps.
 2012-10-01 23:43:47 +10:00
Jan Schmidt
3054545dad resindvd: Port parsetter element 
Port the DVD element which overrides the MPEG encoded pixel-aspect-ratio
to match that requested by the DVD VM.
 2012-10-01 23:28:07 +10:00
Christophe Fergeau
5e544c1556 chromaprint: Missing unmap buffer in transform_ip 
When no samples have to be processed, transform_ip exits early, but
the buffer it uses was not properly unmapped.

https://bugzilla.gnome.org/show_bug.cgi?id=685159
 2012-09-30 20:16:54 +01:00
Jan Schmidt
3c1a1dfec5 resindvd: Fiddle with segment handling for DVD. 
Clarify some of the terms used, and make sure to extend the
stop time when the position moves past the end of the current
segment (as for still frames, for example).

Also, make sure to reset the position value of the segment the
demuxer outputs when collecting a seamless segment.
 2012-09-30 04:08:50 +10:00
Tim-Philipp Müller
4712247e4f Remove some liboil references 
https://bugzilla.gnome.org/show_bug.cgi?id=673285
 2012-09-29 12:36:08 +01:00
Tim-Philipp Müller
346de28846 cog: remove some old elements so they don't get ported accidentally 
They are replaced by videoconvert and gdkpixbufoverlay
 2012-09-29 12:30:20 +01:00
Руслан Ижбулатов
2c2d6b0753 curl: fix compiler warning with setsockopt() on windows 
The optval argument is of type const char * on windows.

https://bugzilla.gnome.org/show_bug.cgi?id=685065
 2012-09-29 11:41:56 +01:00
Руслан Ижбулатов
75dba03688 curl: fix build on win32 
https://bugzilla.gnome.org/show_bug.cgi?id=684847
 2012-09-28 09:57:53 +01:00
Jan Schmidt
7b0e5f4598 soundtouch: Transform timestamps properly after a seek. 
The segment is transformed to match the pitch conversion
being applied, so make sure the timestamps being output
match the configured downstream segment accordingly, and
adjust the downstream segment position to match the stream time
ratio also.
 2012-09-26 14:57:54 +10:00
Jan Schmidt
34510b71ca resindvd: Don't send spurious 'close segment' events. Make a single segment 
With the new 1.0 segment semantics, there's no need to close the
previous segment before sending out a new one, so don't do it.

Also, adjust the outgoing segment to just be one contiguous
timeline, with a base of 0. Fixes some repeating timestamps when
segments change.
 2012-09-26 14:57:54 +10:00
Yaakov Selkowitz
33a012219d chromaprint: link against libgstbase 
Calls to gst_base_transform_get_type and gst_base_transform_set_passthrough
require linking with libgstbase.

https://bugzilla.gnome.org/show_bug.cgi?id=684740
 2012-09-24 23:23:25 +01:00
Tim-Philipp Müller
cc8bd6af0b assrender: disable for autoplugging until it works 
https://bugzilla.gnome.org/show_bug.cgi?id=683192
https://bugzilla.gnome.org/show_bug.cgi?id=684683
 2012-09-23 21:08:30 +01:00
Olivier Crête
58b4ed1949 opusenc: Rank as Primary 2012-09-22 14:58:52 -04:00
Olivier Crête
df0525de11 rtpopuspay: Allocate the rtp buffer correctly 
Use the right functions to allocate the rtp buffer
 2012-09-22 14:58:52 -04:00
Jan Schmidt
c90cdadbe4 resindvd: Fix segment handling in the demuxer. 
Transfer the incoming segment base time to the src
segment the demuxer sends downstream. Add some
(disabled) debug to the bin for tracking outgoing
timestamps.
 2012-09-18 00:27:22 +10:00
Sebastian Dröge
1419862c50 vp8: Remove plugin, moved to gst-plugins-good 2012-09-16 15:48:02 +02:00
Sebastian Dröge
12b5ef6573 vp8dec: Unref input/output states when stopping the decoder 2012-09-16 15:25:08 +02:00
Sebastian Dröge
50ad4b5bbe vp8enc: Update realtime profile to the new properties 2012-09-16 15:21:33 +02:00
Sebastian Dröge
a2503e91bf vp8: Require latest libvpx release (1.1.0 from May 2012) 
Fixes bug #684116 and simplifies configure checks.
 2012-09-16 10:58:29 +02:00
Sebastian Dröge
829c1dcdaf vp8enc: Use a string field for the profile in the caps 
Just for consistency with all the other codecs.
 2012-09-15 20:24:00 +02:00
Mark Nauwelaerts
5fad6e86d7 use gst_element_factory_get_metadata to replace obsolete API 2012-09-15 19:13:47 +02:00
Sebastian Dröge
8da71fea1f vp8enc: Correctly set profile in caps 2012-09-15 00:04:07 +02:00
Sebastian Dröge
f7e0bc0c6a vp8: Update copyright and authors 2012-09-14 23:41:48 +02:00
Sebastian Dröge
c371606bc1 vp8enc: Rework encoder properties to be more in line with the libvpx tools and API 
Also add all available properties.
 2012-09-14 23:37:39 +02:00
Mark Nauwelaerts
578861abea replace gst_element_class_set_details_simple with gst_element_class_set_metadata 2012-09-14 17:27:49 +02:00
Wim Taymans
b8f913555d ffmpegcolorspace is no more 2012-09-14 16:45:34 +02:00
Wim Taymans
0d8f8a5134 rename some caps and elements in examples 2012-09-14 16:29:23 +02:00
Jan Schmidt
16f5120d96 resindvd: Start with a larger subpicture queue for preroll. 
Shrink the queue once we're prerolled. Fixes the Fringe S1
DVD. Also, ensure we only fire no_more_pads once.
 2012-09-12 23:05:47 -07:00
Jan Schmidt
c5e0bc6464 resindvd: Disable input-selector syncing 
Until segments are better worked out. Also, use
internal input-selector for audio as well
 2012-09-12 23:05:47 -07:00
Tim-Philipp Müller
d95620ccd1 flitetestsrc: fix caps fixation 
Make caps writable before changing them.
 2012-09-13 01:34:45 +01:00
Tim-Philipp Müller
4ed3d60bd2 waylandsink: fail gracefully with an error message if we can't connect to wayland 
g_return_val_if_fail() is not for error handling, it's for
catching programming errors in public API.

Fixes problem with generic/states unit test.
 2012-09-13 01:09:56 +01:00
Tim-Philipp Müller
86a8018119 schroenc: remove GST_ERROR statement that was committed by accident 2012-09-13 00:50:04 +01:00
Peter Korsgaard
b04e61d8a4 opus + jpegformat: unbreak non-debug build 
opus + jpegformat plugin builds fail when gstreamer is configured with
--disable-gst-debug as they are checking the GST_DISABLE_DEBUG symbol
instead of GST_DISABLE_GST_DEBUG.

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>

https://bugzilla.gnome.org/show_bug.cgi?id=683850
 2012-09-12 09:13:30 +01:00
Olivier Crête
9b559160cc spanplc: Don't unref too much, don't forward gap event 2012-09-11 20:30:55 -04:00
Olivier Crête
f9320170cd spanplc: Port to 1.0 API 2012-09-11 20:19:48 -04:00
Mark Nauwelaerts
b807753453 ext, gst: only activate in pull mode if upstream is seekable 2012-09-11 17:47:33 +02:00
Mark Nauwelaerts
2816380e7c opusenc: port to the new GLib thread API 2012-09-11 15:12:55 +02:00
Tim-Philipp Müller
1ef529601b opencv: disable motioncells "date" property for now 
Shouldn't be of type long. Not sure it's useful at all, so
just disable it for now to see if someone misses it.
 2012-09-09 23:52:41 +01:00
Rico Tzschichholz
b13b59014c mythtv: update for duration message API change 2012-09-03 19:25:52 +01:00
Tim-Philipp Müller
08149652a1 schroenc: set interlace_coding based on input caps 
and remove property.
 2012-09-03 14:21:19 +01:00
Tim-Philipp Müller
bc93340fd4 schroenc: configure profile/level from allowed downstream caps instead of properties 
This is how it's done elsewhere and works better with encodebin.

Removes "force-profile" property, read-only "profile" property and
"level" property.

https://bugzilla.gnome.org/show_bug.cgi?id=670608
 2012-09-03 14:04:40 +01:00
Tim-Philipp Müller
3d3332f84c schroenc: prettify enum property nicks 
Use '-' instead of '_' as word separator.
 2012-09-03 13:04:01 +01:00
Sreerenj Balachandran
18c5a0f60f spc: No need to define GLIB_DISABLE_DEPRECATION_WARNINGS 
https://bugzilla.gnome.org/show_bug.cgi?id=683175
 2012-09-02 20:01:38 +01:00
Tim-Philipp Müller
a79edf27d0 gst_message_new_duration -> gst_message_new_duration_changed 2012-09-02 01:48:12 +01:00
Jan Schmidt
f2d0c6a89e resindvd/mpegdemux: Send both PTS and DTS downstream 
Now that we can.
 2012-09-01 16:09:59 -07:00
Tim-Philipp Müller
f3e66bd80d spc: don't leak buffer in case of an init error 2012-09-02 00:04:18 +01:00
Sreerenj Balachandran
1d384422e0 spc: port to 0.11 
https://bugzilla.gnome.org/show_bug.cgi?id=683175
 2012-09-02 00:02:14 +01:00
Jan Schmidt
8f719fea02 resindvd: Plug a video parser in front of decoder 
Seems to make playback both better and worse. The parser needs some
fixing, it seems :-/
 2012-08-31 14:22:59 -07:00
Jan Schmidt
211828979b resindvd: Add some GAP event stuff to make still-frames start to work 
And remove the dead audiomunge element. It isn't needed now, we just
send GAP events
 2012-08-31 13:12:41 -07:00
Jan Schmidt
1218cff3dc resindvd: Initial partial port to 0.11 2012-08-31 12:46:29 -07:00
Sreerenj Balachandran
547d679e9e wayland: use scale fullscreen method. 
This makes the compositor scale the surface to fit and preserves
aspect ratio.

https://bugzilla.gnome.org/show_bug.cgi?id=681453
 2012-08-21 10:41:31 +01:00
Sreerenj Balachandran
dc81682f4b wayland: respond to ping/pong protocol so we're not deemed unresponsive. 
https://bugzilla.gnome.org/show_bug.cgi?id=681453
 2012-08-21 10:41:24 +01:00
Sreerenj Balachandran
c803ca4af5 wayland: More fixes based on wayland-0.95.0 and gstreamer-1.0 apis 
- bump wayland version to 0.95.0 which will lead to stable 1.0 release
- avoid memcopy and use propose_allocation for GstBufferPool allocation
- using WaylandBufferPool
- shm: Allocate shm buffers through new wl_shm_pool interface
  (the shm buffer allocation is a two step process now: first
  allocate a wl_shm_pool, then allocate a buffer from the pool)

https://bugzilla.gnome.org/show_bug.cgi?id=681453
 2012-08-21 10:24:45 +01:00
Tim-Philipp Müller
2483c82bd7 video/x-dvd-subpicture -> subpicture/x-dvd 2012-08-20 23:37:00 +01:00
Tim-Philipp Müller
cc167a237a Fix up FSF address in some source files as well 2012-08-18 20:31:17 +01:00
Edward Hervey
508e4e7e9f vp8dec: Call gst_video_decoder_negotiate() 2012-08-14 19:07:47 +02:00
Edward Hervey
54bfdebd9f schrodec: Call gst_video_decoder_negotiate() 2012-08-14 19:07:43 +02:00
Edward Hervey
c4ccca8795 soundtouch: Don't assume output buffer timestamps 
There's no guarantee whatsoever that the first buffer to output will
start at the segment.start.

Instead, wait for the first buffer after a segment, and use that timestamp
 2012-08-14 19:04:42 +02:00
Sebastian Dröge
c6b16aed08 vp8dec: Add support for multiple decoding threads 2012-08-14 11:17:25 +02:00
Sebastian Dröge
14317a3408 vp8dec: Add support for the MFQE postprocessing flag 
Which is enabled by default if postprocessing is enabled.
 2012-08-14 11:09:46 +02:00
Sebastian Dröge
85e45a09e2 vp8: Use pkg-config file for getting the LIBS and CFLAGS 2012-08-09 13:37:22 +02:00
Sebastian Dröge
7369d28e1f mpg123: Give MARGINAL rank to the mpg123 decoder element 2012-08-09 11:48:39 +02:00
Sebastian Dröge
880359f32b vp8enc: Update the per-component strides for every frame too 
This is necessary because of GstVideoAlignment
 2012-08-08 17:41:42 +02:00
Tim-Philipp Müller
7960591c01 mpg123: dist header file 2012-08-08 12:58:50 +01:00
Olivier Crête
bc6b513620 mimic: Port to 0.11 2012-08-07 15:41:25 -07:00
Oleksij Rempel
94337d05a5 vp8enc: initiate encoder to fix a crash. 
Without this patch vp8enc send header before and after first
key frame. On second keyframe vp8dec will crash without getting
decoded frame. With this pipe it is easy to reproduce this issue:
gst-launch-1.0 videotestsrc ! vp8enc ! vp8dec ! fakesink

https://bugzilla.gnome.org/show_bug.cgi?id=680667
 2012-08-05 19:35:32 +01:00
Tim-Philipp Müller
489b102a43 gst_tag_list_free -> gst_tag_list_unref 2012-08-04 16:31:30 +01:00
Patricia Muscalu
a7037cd0e3 curl: fixed some memory leaks in the plugin and the unit tests 
https://bugzilla.gnome.org/show_bug.cgi?id=681117
https://bugzilla.gnome.org/show_bug.cgi?id=681115
 2012-08-03 20:28:40 +01:00
Tim-Philipp Müller
6cac2bff68 mpg123: map input buffer in READ mode, not WRITE mode 
Makes things actually work.
 2012-08-03 13:43:31 +01:00
Tim-Philipp Müller
e7a5cfec30 mpg123: query supported output formats at run-time 
Fixes stuff. We use a string here since we can't be bothered
with GValue.
 2012-08-03 13:38:05 +01:00
Tim-Philipp Müller
f75e4fbc11 mpg123: hook up to build system 2012-08-03 11:52:49 +01:00
Carlos Rafael Giani
da622c27ba mpg123: add new libmpg123-based mp3 decoder plugin 
Needs a bit of cleaning up.

https://bugzilla.gnome.org/show_bug.cgi?id=681003
 2012-08-03 11:52:11 +01:00
Tim-Philipp Müller
c031f5bd27 schroedinger, mpegtsmux: fix plugin names 2012-08-03 11:52:11 +01:00
Mark Nauwelaerts
19ca77e95e faad: do not access frame buffer following _finish_frame call 
Fixes https://bugzilla.gnome.org/show_bug.cgi?id=680494
 2012-08-02 14:40:55 +02:00
Tim-Philipp Müller
a48cfea4a2 mplex: default to DVD MPEG-2 format instead of Generic MPEG-1 2012-08-01 11:34:08 +01:00
Tim-Philipp Müller
e1e29488d0 mplexjob: add nicks for format enums, add some more formats 2012-08-01 11:31:04 +01:00
Sebastian Dröge
afd50a79f4 tag: Update for taglist/tag event API changes 2012-07-28 00:32:58 +02:00
Mark Nauwelaerts
dd3ae904f9 assrender: process ATTACHMENT tag as sample 2012-07-27 13:22:52 +02:00
Mark Nauwelaerts
f890f60ee2 faac, voaacenc: adjust to modified audioencoder getcaps helper API 2012-07-25 16:05:41 +02:00
Sebastian Dröge
a7ae20305a schro: Add support for video metadata and arbitrary strides 2012-07-23 11:36:01 +02:00
Sebastian Dröge
47546ab2df schro: Map raw video buffers with the correct read/write flags 2012-07-23 10:54:17 +02:00
Sebastian Dröge
b60da6223e ext: Update for video base classes API changes 2012-07-23 10:35:03 +02:00
Oleksij Rempel
c03cc692ef vp8enc: fix memory leak 
unref frame. i hope it is correct place to do it.

Signed-off-by: Oleksij Rempel <bug-track@fisher-privat.net>
 2012-07-23 08:31:11 +02:00
Mark Nauwelaerts
9e28b769e8 faad: only discard frame once upon decoding error 
Fixes https://bugzilla.gnome.org/show_bug.cgi?id=679886
 2012-07-20 16:47:35 +02:00