Just to rephrase the difference between choices #3 and #4:

#3 means to make an entirely new data representation (say, some kind of struct 
with various fields) and have to somehow make TypeDesc, ParamValueList, and 
potentially lots of apps, deal with it.

#4 means to encode the timecode as an existing data type (a string, or a 
uint[2]) -- for which ParamValueList and apps that don't care about timecodes 
can continue to operate as they always have -- but have the 'vecsemantics' 
field (or use the 'reserved' field) in TypeDesc encode that it's supposed to be 
interpreted as a timecode.  Minimally intrusive. But before enshrining "these 
bytes are an SMPTE timecode" in the TypeDesc, I'd want us to be really, really 
sure we agreed on what the encoding is and that we won't discover later that we 
need to change it.



On Oct 10, 2013, at 10:18 AM, Larry Gritz wrote:

> 3. Make an entirely new type, 

> 4. We could transform the 'reserved' field in TypeDesc to some kind of 
> "special meaning" code -- or just make another tag for the existing 
> 'vecsemantics' byte -- and let "this is an SMPTE timecode" be one of those 
> values.  

--
Larry Gritz
[email protected]


_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org

Reply via email to