| Re: [linuxtools-dev] eclipse lttng viewer vs. lttv-gui |
Hello,
The file
format should not
be an issue for the representation of the states, I have tried a couple
of files
in 2.3 format and detected some differences but far less than with your
file.
I.e. in one of the files I tried with 380 processes,
the
only
UNNAMED process was the swapper while lttv was able to resolve the
name.
There were 2 out of 380
processes with a birth time of 0.
We will look into these
differences and
align with lttv,
There
is no specific bug to track the differences as mentioned in your
e-mail below,
Please file a bug so we can track the progress on it.
In
order to have a common ground to
discuss it, please submit the traces. If they are very large please e-mail me directly so I can fetch the logs.
Best
Regards
/Alvaro
Hi,
I recognize that the lttng viewer for eclipse is at the .5 state, but here are some questions that relate to its current functionality. Please let me know if I can find this information out somewhere else (bug database?):
We are loading the same trace file into the eclipse viewer as we are into the LTTV viewer and are seeing differences between them. For instance:
- The time values (both birth and nanos) are 0 for many entries (>90%) in eclipse but they almost all have values in the lttv viewer (The swapper process has 0 values in LTTV).
- The number of events is different across the two. There are many UNNAMED processes in the Eclipse viewer.
We are running on 2.3 traces. Would this be an issue with the trace format?
Thanks,
n Rick
_______________________________________________
linuxtools-dev mailing list
linuxtools-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/linuxtools-dev