Discussion:
TimeTags: "t" Type Tag, or attaching the TimeTag to a bundle? Which is the current correct way? Both?
(too old to reply)
Ilias Bergstrom
2014-08-06 10:08:12 UTC
Permalink
Hi!

I am in the process of finishing a nice big program with OSC at its core,
and I want to support the protocol properly to the greatest extent
possible. I am using Ross's excellent OSCpack library (1.1.0).

I have left TimeTag support to last because I was the most confused by
them, but now that's the only piece missing so I have to tackle it :)

I see two ways of handling TimeTags in the OSC documentation: one, the 1.0
way, required the use of bundles, to which the TimeTag was attached.

In the 1.1 NIME09 paper, the "t" Type Tag is specified for the TimeTag.

Should a modern OSC program support both? Or is the requirement of a
message bundle to support TimeTags deprecated/superseded by the "t" tag?

Thank you!

Ilias Bergstrom
Matt Wright
2014-08-06 16:30:19 UTC
Permalink
The idea is that every bundle has a time tag saying when the receiver should make the bundle take effect. Those are the semantics of the time tags.

Later we realized it would also be useful to allow time tags as arguments to messages, e.g., “it was time X on my clock when I received the bundle you sent that had time-tag Y”, so we added the type “t” to provide a way to send this particular kind of 64-bit fixed point number as a message argument. That part is probably easy to implement; just pass over the 64 bits like you would a double.

“Implementing time tags” means the former, ideally building some kind of scheduling queue into your receiver (but in practice often much less).

-Matt
Hi!
I am in the process of finishing a nice big program with OSC at its core, and I want to support the protocol properly to the greatest extent possible. I am using Ross's excellent OSCpack library (1.1.0).
I have left TimeTag support to last because I was the most confused by them, but now that's the only piece missing so I have to tackle it :)
I see two ways of handling TimeTags in the OSC documentation: one, the 1.0 way, required the use of bundles, to which the TimeTag was attached.
In the 1.1 NIME09 paper, the "t" Type Tag is specified for the TimeTag.
Should a modern OSC program support both? Or is the requirement of a message bundle to support TimeTags deprecated/superseded by the "t" tag?
Thank you!
Ilias Bergstrom
_______________________________________________
OSC_dev mailing list
http://lists.create.ucsb.edu/mailman/listinfo/osc_dev
Ilias Bergstrom
2014-08-07 08:05:25 UTC
Permalink
Hi Matt, thank you!

In the end I spent a day and both bundle and 't' timetags are implemented.
I'll need to test with a program that takes them seriously (I've yet to
come across one, probably because I use OSC for visuals where they matter
less), but at least with the Max OpenSoundControl and oscP5 Processing
objects it all seems to work fine.

I tend to view things I don't know about as being harder than they really
are :)

I'll post back here in the near future when my software is out for Alpha
testing, it is very much around OSC so it may be of interest.

Thanks again!

Best,
Ilias Bergstrom
Post by Matt Wright
The idea is that every bundle has a time tag saying when the receiver
should make the bundle take effect. Those are the semantics of the time
tags.
Later we realized it would also be useful to allow time tags as arguments
to messages, e.g., “it was time X on my clock when I received the bundle
you sent that had time-tag Y”, so we added the type “t” to provide a way to
send this particular kind of 64-bit fixed point number as a message
argument. That part is probably easy to implement; just pass over the 64
bits like you would a double.
“Implementing time tags” means the former, ideally building some kind of
scheduling queue into your receiver (but in practice often much less).
-Matt
Post by Ilias Bergstrom
Hi!
I am in the process of finishing a nice big program with OSC at its
core, and I want to support the protocol properly to the greatest extent
possible. I am using Ross's excellent OSCpack library (1.1.0).
Post by Ilias Bergstrom
I have left TimeTag support to last because I was the most confused by
them, but now that's the only piece missing so I have to tackle it :)
Post by Ilias Bergstrom
I see two ways of handling TimeTags in the OSC documentation: one, the
1.0 way, required the use of bundles, to which the TimeTag was attached.
Post by Ilias Bergstrom
In the 1.1 NIME09 paper, the "t" Type Tag is specified for the TimeTag.
Should a modern OSC program support both? Or is the requirement of a
message bundle to support TimeTags deprecated/superseded by the "t" tag?
Post by Ilias Bergstrom
Thank you!
Ilias Bergstrom
_______________________________________________
OSC_dev mailing list
http://lists.create.ucsb.edu/mailman/listinfo/osc_dev
_______________________________________________
OSC_dev mailing list
http://lists.create.ucsb.edu/mailman/listinfo/osc_dev
Loading...