[OSC_dev] blob metadata (was: Images ?)

[salsaman at xs4all.nl]

Huh...? This makes no sense at all to me - the "C" in OSC stands for
*control* - in other words it is a control channel, not a data channel...!

To me it is far more logical to send any data out of band, but for example
you could enclose this with a "sending data" / "data-received","checksum"
exchange.

This is exactly what I am doing with LiVES - OSC is control only, actual
video and audio data are sent via other routes (e.g. video/audio jack,
ogg)

There are various advantages to splitting this way, for example you can
more easily interlace/multiplex your data, you can send control messages
during the data transfer (for example a restart/cancel, volume change,
etc), also the command channels are kept quieter, thus more responsive.

Seems a bit lazy to me to want to send control and data all on the same
channel.


Just my opinion...

Salsaman.
http://lives.sourceforge.net



On Thu, March 5, 2009 19:33, Stephen Sinclair wrote:
> The idea occurred to me once that it certainly would be nice to be
> able to send meta data with blobs so that they could be interpreted
> correctly.  This could include applications like streaming audio data
> -- you would want information about sample rate, compression, etc.
> Another use case for meta data would be for sending things like
> matrices: row/column information would be necessary.  (Right now I am
> sending matrices as lists of 9 or 16 floats)
>
> Well, for the former case (images, sound files), the internet is
> fortunately already well-populated with standards for file types:
> e.g, http://en.wikipedia.org/wiki/MIME
>
> I suppose you could even just send a message with two blobs: a MIME
> header and a file.  Of course, it might not fit in a UDP message, so
> you might want to use TCP, and in that case, perhaps you also might as
> well be using HTTP instead of OSC... but I digress.
>
> For the problem of sending metadata about vector/matrix formats, I
> have no solution except to define application-specific standards.
> (i.e., my receiver "knows" that a certain message has a 3x3 matrix as
> its 9 arguments.)  Does anyone know if there are any existing metadata
> standards for identifying mathematical data structures?
>
>
> Steve
>
>
> On Thu, Mar 5, 2009 at 3:07 AM, Gaspard Bucher <gaspard at teti.ch> wrote%3