[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
[linuxtools-dev] [lttng] CTFTraceReader inconsistencies
|
Hi,
There seems to exists inconsistency in CTFTraceReader timestamps.
* getStartTime() returns a raw timestamps
* A call to getEndTime() after object instantiation returns a raw
timestamps equals to start.
* After reading an event, getEndTime() returns the timestamps + offset
of the trace.
When start has no offset, this leads to a funny result: duration = end -
start = ~42 years (2012 - 1970)
Also, a seek should be performed with a timestamps relative to start,
not (start + offset). Thus, reader.seek(reader.getEndTime()) if done
after object instantiation returns the first event in the trace, and if
done after goToLastEvent(), returns a null event, because of the offset.
I would suggests that timestamps getters should all adds offset or not,
and that seek should follow the same convention by accepting timestamps
returned by getStartTime() and getEndTime().
On a side note: goToLastEvent() is very not efficient, because all
events in the trace is processed to reach the end. Could it be possible
to directly jump to the end instead?
Cheers,
Francis
Attachment:
smime.p7s
Description: Signature cryptographique S/MIME