Re: minutes of 2010-10-06 teleconference

Sorry I was travelling during the meeting. My opinion below...

2010/10/6 Raphaël Troncy <raphael.troncy@eurecom.fr>

<...>


>   2. MEDIA FRAGMENT EXTENSION
>
>   <raphael> close ACTION-188
>
>   <trackbot> ACTION-188 Remind Jack and Michael to register asap for
>   TPAC closed
>
>   <raphael> Raphael: summarizing the long thread and discussion about
>   extensibility
>
>   <raphael> ... and summarizing the position of philip
>
>   <raphael> Jack: the only thing we should say is: if we partially
>   understand a media fragment, should we accept or ignore ?
>
>   <raphael> ... this is one line in the spec
>
>   <raphael> ... we should accept what is acceptable
>
>   <raphael> ... and in particular if a media fragment URI can have
>   things we don't understand
>
>   <raphael> ... and personally I don't care about the grammar
>
>   <raphael> Yves: we should be very caution with extension
>
>   Yves: we should be careful with extensions
>
>   <raphael> ... e.g. the ref redefines an existing dimension
>
>   Jack: explains we could use an attribute to control strict or loose
>   parsing
>
>   All: think this is ugly
>
>   <Yves> :)
>
>   Jack: suggestion: we mandate the chair to decide something
>
>   <raphael> Jack: I have no strong feeling either way
>
>   Jack: I do think (and so does Davy) that if we recognize the
>   attribute name then the value must adhere to our syntax
>
>   <raphael> #t=banana ... means parsing stops
>
>   <davy> #t=10&t=20 ... means parsing stops?
>
>   <Yves> well if you always ignore what you don't understand
>   #t=banana&xywh=10,10,20,20 should be xywh=10,10,20,20
>


I agree with Yves: the first name-value pair is invalid and should be
ignored, but the second name-value pair is valid and should be interpreted.



>   <davy> I disagree
>
>   That woulnd mean that t=1o&xywh=1,2,3,4 would parse
>


Yes, but only the second name-value pair.


  <Yves> this is what is consistent with 'ignoring what you don't
>   understand'
>
>   <Yves> we have t=npt:120, what happens if we have t=<mytimeunit>:120
>   ?
>
>   <Yves> we can have extensionaxis = extensionname = extensionvalue
>
>   <raphael> Jack: we should not try to encode in a grammar what we
>   will not understand
>
>   <Yves> but jack, encoding extension points is a good thing to do
>
>   Yves: either we do nothing about extension or do it correct, with
>   extensionaxis in the grammar
>

Yes, I agree, it should be in the grammar. Maybe that can alleviate Philip's
concern and make it easy to include the extensibility.



>   ... we could check what existing implementations do
>
>   Raphael: they are waiting for us to specify correct behaviour
>   ... time to adjourn
>
>   <raphael> Jack: I suggest the chairs put a deadline and take a
>   decision on this issue
>
>   <raphael> Raphael: I agree, I will have a discussion with Erik
>
> Summary of Action Items
>
>   [End of minutes]
>     _________________________________________________________
>
>
Cheers,
Silvia.