[gst-devel] audio/raw vs float

Simon Per Soren Kagedal simon at cs.uoregon.edu
Tue Mar 20 01:11:51 CET 2001


On Mon, Mar 19, 2001 at 10:25:58PM +0000, Richard Boulton wrote:
> I don't think theres actually a need for this.  The caps are to define
> what endianness an element is _able_ to cope with.  If the element
> can cope with either endianness, it should just specify that both values
> are acceptable (or not specify the property at all, if its certain that
> there will never be any other possible values).

Sure, I agree with that (although I can't see why an element would
want to accept both kinds of endianness, except maybe for the sole
purpose of converting the stream to host's endianness).  I'm just
saying that in 99% of cases, elements will want to operate on
host-endian data only, and it should be easy to specify that.

But I'm not completely sure what we're talking about here, I have to
admit.  Is the current (= gstreamer 0.1.1, for me) underlying
framework of caps and metadata basically going to be kept, or is this
something completely new?

Btw, on the question about other endiannesses...  there is PDP-endian
:) but we will never encounter it and it doesn't affects 16-bit ints
anyway.

Simon.




More information about the gstreamer-devel mailing list