This is something I stumbled upon while looking at string copying we do (in spades) and did a simple change to not be constantly copying things around needlessly. There's a lot more that can be done in these files of course so this is sort of a reminder for future code edits that it's possible to design interfaces/function in a way that's more performance aware and avoid forcing creation of copies, while still being very simple. Also, we can use lambdas now :)
BUG=
R=perkj@webrtc.org, pthatcher@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/41589004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8131 4adac7df-926f-26a2-2b94-8c16560cd09d
The purpose of this CL is to add a new file in libjingle without breaking Chromium in the process. The plan is to do the following:
1. Land a no-op videoframefactory.cc in webrtc (this file).
2. Wait for it to roll into Chromium.
3. Modify libjingle.gyp in Chromium to include this file.
4. Make the real change in webrtc with the real implementation of this file.
5. Wait for the change to roll into Chromium.
R=tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/28049004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7712 4adac7df-926f-26a2-2b94-8c16560cd09d
This didn't compile on the FYI bots. Example error:
FAILED: E:\b\depot_tools\python276_bin\python.exe gyp-win-tool link-with-manifests environment.x86 True chrome_child.dll "E:\b\depot_tools\python276_bin\python.exe gyp-win-tool link-wrapper environment.x86 False link.exe /nologo /IMPLIB:chrome_child.dll.lib /DLL /OUT:chrome_child.dll @chrome_child.dll.rsp" 2 mt.exe rc.exe "obj\chrome\chrome_child_dll.chrome_child.dll.intermediate.manifest" obj\chrome\chrome_child_dll.chrome_child.dll.generated.manifest
content_renderer.lib(content_renderer.webrtc_video_capturer_adapter.obj) : error LNK2001: unresolved external symbol "public: virtual class cricket::VideoFrame * __thiscall cricket::VideoFrameFactory::CreateAliasedFrame(struct cricket::CapturedFrame const *,int,int,int,int)const " (?CreateAliasedFrame@VideoFrameFactory@cricket@@UBEPAVVideoFrame@2@PBUCapturedFrame@2@HHHH@Z)
libjingle_webrtc_common.lib(libjingle_webrtc_common.peerconnectionfactory.obj) : error LNK2001: unresolved external symbol "public: virtual class cricket::VideoFrame * __thiscall cricket::VideoFrameFactory::CreateAliasedFrame(struct cricket::CapturedFrame const *,int,int,int,int)const " (?CreateAliasedFrame@VideoFrameFactory@cricket@@UBEPAVVideoFrame@2@PBUCapturedFrame@2@HHHH@Z)
libjingle_webrtc_common.lib(libjingle_webrtc_common.videocapturer.obj) : error LNK2001: unresolved external symbol "public: virtual class cricket::VideoFrame * __thiscall cricket::VideoFrameFactory::CreateAliasedFrame(struct cricket::CapturedFrame const *,int,int,int,int)const " (?CreateAliasedFrame@VideoFrameFactory@cricket@@UBEPAVVideoFrame@2@PBUCapturedFrame@2@HHHH@Z)
libjingle_webrtc_common.lib(libjingle_webrtc_common.dummydevicemanager.obj) : error LNK2001: unresolved external symbol "public: virtual class cricket::VideoFrame * __thiscall cricket::VideoFrameFactory::CreateAliasedFrame(struct cricket::CapturedFrame const *,int,int,int,int)const " (?CreateAliasedFrame@VideoFrameFactory@cricket@@UBEPAVVideoFrame@2@PBUCapturedFrame@2@HHHH@Z)
chrome_child.dll : fatal error LNK1120: 1 unresolved externals
> cricket::VideoAdapter: Drop frames before spending time converting/scaling, not after.
>
> In VideoCapturer::OnFrameCaptured, we currently convert cricket::CapturedFrame to cricket::VideoFrame and then send that to VideoAdapter::AdaptFrame. AdaptFrame may then decide to drop the frame. It would be faster to drop the frame before converting to cricket::VideoFrame.
>
> This CL refactors VideoAdapter with a new function AdaptFrameResolution that takes captured resolution as input and output adapted resolution, or 0x0 if the frame should be dropped. Using that function, frames can be dropped before any conversion takes place.
>
> R=fbarchard@google.com, perkj@webrtc.org, tommi@webrtc.org
>
> Committed: https://code.google.com/p/webrtc/source/detail?r=7702
>
> Review URL: https://webrtc-codereview.appspot.com/29949004TBR=magjed@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/28019004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7708 4adac7df-926f-26a2-2b94-8c16560cd09d
In VideoCapturer::OnFrameCaptured, we currently convert cricket::CapturedFrame to cricket::VideoFrame and then send that to VideoAdapter::AdaptFrame. AdaptFrame may then decide to drop the frame. It would be faster to drop the frame before converting to cricket::VideoFrame.
This CL refactors VideoAdapter with a new function AdaptFrameResolution that takes captured resolution as input and output adapted resolution, or 0x0 if the frame should be dropped. Using that function, frames can be dropped before any conversion takes place.
R=fbarchard@google.com, perkj@webrtc.org, tommi@webrtc.org
Committed: https://code.google.com/p/webrtc/source/detail?r=7702
Review URL: https://webrtc-codereview.appspot.com/29949004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7707 4adac7df-926f-26a2-2b94-8c16560cd09d
Rease for revert: failed internal test cases
> cricket::VideoAdapter: Drop frames before spending time converting/scaling, not after.
>
> In VideoCapturer::OnFrameCaptured, we currently convert cricket::CapturedFrame to cricket::VideoFrame and then send that to VideoAdapter::AdaptFrame. AdaptFrame may then decide to drop the frame. It would be faster to drop the frame before converting to cricket::VideoFrame.
>
> This CL refactors VideoAdapter with a new function AdaptFrameResolution that takes captured resolution as input and output adapted resolution, or 0x0 if the frame should be dropped. Using that function, frames can be dropped before any conversion takes place.
>
> R=fbarchard@google.com, perkj@webrtc.org, tommi@webrtc.org
>
> Review URL: https://webrtc-codereview.appspot.com/29949004TBR=magjed@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/24279004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7703 4adac7df-926f-26a2-2b94-8c16560cd09d
In VideoCapturer::OnFrameCaptured, we currently convert cricket::CapturedFrame to cricket::VideoFrame and then send that to VideoAdapter::AdaptFrame. AdaptFrame may then decide to drop the frame. It would be faster to drop the frame before converting to cricket::VideoFrame.
This CL refactors VideoAdapter with a new function AdaptFrameResolution that takes captured resolution as input and output adapted resolution, or 0x0 if the frame should be dropped. Using that function, frames can be dropped before any conversion takes place.
R=fbarchard@google.com, perkj@webrtc.org, tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/29949004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7702 4adac7df-926f-26a2-2b94-8c16560cd09d
Allows successful build of arm64 libraries using
GYP_DEFINES="OS=ios target_arch=arm64 target_subarch=arm64".
Note that not all libraries will be NEON optimized (eg common_audio),
however most importantly libvpx will be. WEBRTC_ARCH_ARM needs to be
defined so that libvpx doesn't post-process, which is significantly
detrimental to performance.
BUG=3898
R=kjellander@webrtc.org, pthatcher@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/26959004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7573 4adac7df-926f-26a2-2b94-8c16560cd09d
Removes direct VideoCodec use from the new API, exposes VideoDecoders
through webrtc/video_decoder.h similar to VideoEncoders.
Also includes some preparation for wiring up external decoders in
WebRtcVideoEngine2 by adding AllocatedDecoders that specify whether they
were allocated internally or externally.
Additionally addresses a data race in VideoReceiver that was exposed with this change.
R=mflodman@webrtc.org, stefan@webrtc.orgTBR=pthatcher@webrtc.org
BUG=2854,1667
Review URL: https://webrtc-codereview.appspot.com/27829004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@7560 4adac7df-926f-26a2-2b94-8c16560cd09d
