I think this would be a valuable addition to Chrome. Please consider adding
streaming
support via RTSP and RTP to chrome. This will truly begin to "solve the
problem" of
including video without plugins.
--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings
I also think it's a good idea to support rtsp:///xx.xx.xx.xx/abc.mpg in
chrome
browser, now the html5 video tag can only support http:// ftp:// file://
video
We do some cool tricks with HTTP to allow random access seeking via range
requests.
It's not as nifty as RTSP but Chromium rarely ever downloads the whole
video and
relies on "streaming" audio/video over HTTP.
Lack of live streaming support is the only thing holding HTML5 from getting
flash out
of the picture. I'd suggest adopting apple's support on the iPhone for .m3u8
playlists with streaming audio and video items. You can do stuff like this:
http://www.ioncannon.net/programming/452/iphone-http-streaming-with-ffmpeg-and-an-open-source-segmenter/
Oh I've looked at that and would definitely prefer an HTTP solution over
supporting a
whole new protocol.
The unfortunate part is the only client-side implementation is in
QuickTime. I'm not
aware of any open source implementation.. it's not that it's a very
challenging spec
to implement, but more a matter of finding the time :\
Can anyone tell me if comment 14 is correct? I am trying to do exactly
that, without success. Should Content-Length be -1, or 0, or absent?
Content-Length should be absent. I have used an Icecast2 video source as
the src="" parameter of <video>, which works fairly reliably in Firefox,
but is still a little flaky in Chrome. Above all, Firefox is capable of
timeshifting within a live stream, whereas Chrome is not.
This bug is for RTSP support. I have created Issue 54198 - Support HTTP
Live Streaming. (I'm surprised it didn't exist already!)
Regarding comment #14, the ubiquity of NATs doesn't affect this case except
for in particularly locked-down networks, typically within corporate
environments. The reason is the NAT is most often only on the client side,
and most NATs will properly map incoming UDP packets to clients that have
previously sent outgoing ones. As long as the setup for the RTP stream
first sends outgoing UDP packets to the server (typically using RTSP), the
stream should be successfully established. The cases where it can't are
typically networks that do some global UDP blocking, and you don't have
much chance their in any scenario. The cases where NATs really get tricky
are when both sides are NATted, but that's atypical for the browser-based
use case.
I would love this feature was supported
This will be very nice!