[MPlayer-dev-eng] [PATCH] libmpdemux/realrtsp handling url parameters (start|end)=hh:mm:ss.msec

[rgselk]

--- Roberto Togni <r_togni at tiscali.it> wrote:
> On 2003.11.17 14:54, rgselk wrote:
> > --- Attila Kinali <kinali at gmx.net> wrote:
> > > On Mon, 17 Nov 2003 02:32:29 -0800 (PST)
> > > rgselk <rgselknospam at yahoo.com> wrote:
> [...]
> > The first is a patch to use the parameters 'start' and 'end' in the
> > url
> > to set the rtsp 'Range' parameter.
> > Last Friday I discovered that realplay still uses the complete url
> in
> > the request, so I changed the patch to do the same.
> How does this patch relate to the patch proposed in thread  
>
http://www.MPlayerHQ.hu/pipermail/mplayer-dev-eng/2003-August/019928.html

My patch is more generic. My last version sent here does not remove the
parameters from the url anymore and it handles an end tag as well. I
sniffed some sessions with realplay and realplay uses the same 'Range'
parameter.

> At that time it was not applied. Tha author claimed that the server
> did  
> not like the part following the ? in the url; in his case he had ""  
> around the time values (urls coming from .ram files).
> Maybe we should remove them if coming from ram files (or if they have
>  
> "") and keep them in other cases?

My patch does not handle quotes yet, but that should not be very
difficult to add. I am only testing with 'omroep.nl' which is a helix
server.

> In another mail from that thread Ross from live.com said that clients
>  
> should not parse query strings in urls, because it's the server that 
> 
> have to do it; but maybe with Real proprietary protocol that's not  
> true.

realplay does the same.

> 
> Do you have any idea on the matter?
> 
> BTW, these patches and your other ones are identical (the only  
> difference is that  two debuf printf are commented out):it this the  
> right version or do you still have to send the modified patch that
> does not remove the urls?

You should have used diff to get the other (important) change :-)
One line up now reads:
+    s->param++;
instead of:
+    *(s->param++) = '\0';

> > 
> > The second patch is to find the string .rm or .ra followed by a '?'
>  
> > or
> > '&' or at the end of the string. This allows for detecting a real  
> > rtsp
> > url even when there is another '.' in the string. (for example in
> the
> > millisecond part of the start tag: hh:mm:ss.msec).
> This part should be applied anyway, else those urls won't be
> recognized  
> as Real streams.
> 
> [...]
> I see you're now working on Real rtsp keyframes and seeking: is it  
> still ok to apply this patch or will it be obsoleted by your future  
> patches?

The patches are independant of each other. The only thing is I used the
start tag to discover the bit representing the keyframe. (It is always
in the first packet sent by the server: change the start time and watch
the resulting first packet ;-)

> Ciao,
>  Roberto

Regards,
  rgselk.



__________________________________
Do you Yahoo!?
Free Pop-Up Blocker - Get it now
http://companion.yahoo.com/