Re: libmng data-push mechanism(s)

"Gerard Juyn" <gjuyn-qWit8jRvyhVmR6Xm/[email protected]> Wed, 14 Apr 2004 19:42:42 +0200
Newsgroups gmane.comp.graphics.mng.general
Organization Triple-T
Message-ID <407D9432.26155.531ADE3@localhost>
Ok, so I don't like it when things don't work, so....

Just updated CVS with a working version of the data-push mechanism. 
(anonCVS will take some time to catch up as usual).

This works with a little test-app that pushes data in at around 20ms intervals 
of exactly 379 bytes in size (just for fun...). This means libmng is suspending 
a lot at the start to get the necessary data, but it does work as it's supposed 
to with various sized MNGs.

Unfortunately, contrarary to my earlier contradiction, the suspension mode 
flag must be left off!!! After re-reading the code I can only come to the 
conclusion that it's just inserting a 32Kb buffer in the data-transport which is 
completly unneccesary with the push mechanisms. libmng will still go into 
suspension (even with the flag off) if it receives "0 bytes read" from the 
readdata() callback, and that's exactly what we want.

And so, back to Alex....
Let's see if it works there...

Gerard


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