Re: [Gnash-commit] gnash ChangeLog libbase/FLVParser.cpp libbase/F...

[strk]

On Thu, May 17, 2007 at 01:12:53AM +0200, Tomas Groth wrote:
> 
> --- strk <address@hidden> skrev:
> 
> > On Wed, May 16, 2007 at 09:10:26PM +0200, strk wrote:
> > > On Wed, May 16, 2007 at 06:22:32PM +0000, Tomas Groth wrote:
> > > 
> > > > +       /// Returns the "bufferlength", meaning the differens between 
> > > > the
> > > > +       /// current frames timestamp and the timestamp of the last 
> > > > parseable
> > > > +       /// frame. Returns the difference in milliseconds.
> > > > +       uint32_t getBufferLength();
> > > 
> > > I'm not sure this is what we really need.
> > > If this is what you get by reading NetStream.bufferLength
> > > then I'd expect it to decrease and increase as the stream goes.
> > > In any way I'd expect it to NEVER become == total length if
> > > we play while buffering.
> 
> Is the problem that it increases and decreases? why would that be a problem?
> And no, if we play while buffering it should never return the total length of
> the FLV.

Confirmed. No problem. bufferLength does increase and decrease. See 
NetStream-SquareTest.swf
We'll have to add anothere test there: seeking back. From the comment above 
seeking-to-0
and stopping should result in bufferLength being == total length, we want to 
verify this
(I suspect the buffer stops being filled)

> > Another though.. can NetStream play a continuos stream,
> > like an IP radio ?
> > In this case we can't think the buffer will keep growing...
> > 
> NetStream is suppose to support this, and I think it might already works, but
> because of the reason you mention it is not officially supported yet. We'll
> need to modify the way curl_adaptor stores data, or something like that...

After the release...

--strk;