[FFmpeg-devel] [PATCH] Non-blocking HTTP

[Nicolas George]

Le primidi 11 vent?se, an CCXIX, Luca Barbato a ?crit?:
> That is quite promising but might be generalized soon before or after it
> is applied.
> 
> I'm eventually getting to add some protocol experiment so I could feel
> better how that works with it.

I am not sure it can actually be generalized. Take a poll for writeability
for example: the library can not just poll all the substreams for
writeability since it will return as soon as any of them would become
writeable, not necessarily the one the protocol would actually need to write
to.

(I think this sentence is quite convoluted, so let me rephrase with an
example: imagine a protocol with two substreams: control and payload, poll
will return as soon as control becomes writeable, but the protocol needs to
write on payload, not control.)

Anyway, if you want more time to experiment, there is no rush to apply those
patches.

> PS: url_read_complete is _really_ a pain for seqpacket protocols

Fortunately, unless I missed some places, it is currently only used for HTTP
and TCP.

Regards,

-- 
  Nicolas George
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20110306/0f3080c3/attachment.pgp>