Mandatory name for some field in a CTF trace
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 6 lines · * Yannick Brosseau ([email protected]) wrote:
Hi all,
I'm writing a CTF writer for perf and I'm discovering that some mandatory fields for a CTF trace are not documented.
We must call the time field "timestamp" and the id field mapping event to their description must be "id".
Yes, but they are not mandatory. They can be left out entirely.
Quoted 3 lines
Looking at parser source code, we can find other field like that.
Would it be interesting to add this information in the CTF spec?
I've put these fields as "examples" in the CTF spec, thinking that some would evolve into fixed field names in the future as usage confirms that it's indeed the right choice.
Quoted 4 lines
Or
should we write a second spec on the side for that kind of information? The question we want to answer is "What do I need to put in a CTF file to make it possible for a viewer to display it?"
Beware though: a viewer could view a trace that lacks event ID and timestamp fields entirely. It would be limited to a single even type, and timestamp would always appear as 0.
Thanks,
Mathieu
Quoted 7 lines
Yannick
_______________________________________________
lttng-dev mailing list
[email protected]
http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
Signature · 3 lines
Mathieu Desnoyers EfficiOS Inc. http://www.efficios.com
A message that is not in the archive (sent off-list or not yet received): 20130805194839.GA22120@Krystal
Quoted 9 lines · Mathieu Desnoyers
* Yannick Brosseau ([email protected]) wrote:
[...]Quoted 4 lines
Or
should we write a second spec on the side for that kind of information? The question we want to answer is "What do I need to put in a CTF file to make it possible for a viewer to display it?"Beware though: a viewer could view a trace that lacks event ID and timestamp fields entirely. It would be limited to a single even type, and timestamp would always appear as 0.
On a related note, here's how we dealt with it in TMF:
https://bugs.eclipse.org/bugs/show_bug.cgi?id=387039
Cheers,
Alex