Re: Ogg encapsulation

On Sat, Apr 03, 2004 at 12:01:24PM +0200, Gerard Juyn wrote:

> 0x0002 generate crc in output streams
> (the default then becoming 0x093)

I'd suggest this mean libmng does not calculate the crc, but 
nevertheless writes 4 bytes (e.g. zeros) for that field. It's easy 
enough to stip it out, and the client can fix it up later.

If you want to add another bit for not writing it; I wouldn't object. 
Hmmm. I guess from the abuse point of view, one chooses between parsable 
datastreams with incorrect crcs, and completely unparsable ones. That 
argues for the flag above *not* writing a dummy field.

> Skipping the length would be trivial too as libmng writes chunks at a time, so in 
> the write() callback you'd just skip the first 4 bytes and write length-/-4.
> 
> Convenient enough Ralph? (and thanks for updating autogen.sh!!)

Yep, that certainly makes it easy to leave both out. Thanks! Given all 
this, I think I will reverse my earlier decision and not include the 
length and crc fields in the Ogg MNG packets.

 -r

--
Send the message body "help" to [email protected]