Hi Gerlando,
First the timeGraphView id has to be unique, I'm not sure if the one
you're using is clashing with something else.
If you have only one XML-defined view you might have some luck by
opening LTTng > XML Time Graph View. But all the XML-defined time graph
views compete for this single view, it's not a fully correct
implementation yet...
Normally you would see your XML-defined view under the analysis, under
the trace in the Project Explorer. The view will show up there only
once
the trace is opened. Then you open the view by double-clicking on it in
the Project Explorer.
Patrick
On Mon, Aug 3, 2015 at 12:07 PM, Gerlando Falauto
<gerlando.falauto@xxxxxxxxxxx <mailto:gerlando.falauto@xxxxxxxxxxx>>
wrote:
Hi Patrick, Matthew,
thank you for your kind answers. Sorry for the late reply but I'm
sort of doing this in my spare time.
It all seems quite interesting but I don't exactly know what I'm
looking for, so I'm playing with it one step at a time here.
I somehow managed to start parsing the log files and get some
events
visible in the state system explorer.
One question I actually do have, though: how can I open the time
graph view I defined in the analysis file?
<timeGraphView
id="org.eclipse.linuxtools.tmf.analysis.xml.ui.views.statesystem">
<head>
<analysis id="my.test.state.provider" />
<label value="My Sample XML View" />
</head>
<!-- StateValues -->
<definedValue name="The process is running" value="100"
color="#118811" />
<definedValue name="Critical section" value="101"
color="#881111" />
<definedValue name="Waiting for critical section"
value="102" color="#AEB522" />
<!-- Control Flow View -->
<entry path="Tasks/*">
<display type="self" />
</entry>
</timeGraphView>
After I import it through "Import XML analysis", I was
expecting it
to show up under View / Show View, but I don't. I also checked
that
<analysis id="my.test.state.provider" /> matches <stateProvider
version="0" id="my.test.state.provider">.
Is there any way to enable debug/verbose output when importing XML
file? This all happens in a very (disturbingly) silent way...
And I'm probably missing something stupid...
Thank you so much!
Gerlando
On 07/23/2015 04:27 PM, Patrick Tasse wrote:
Hi Gerlando,
You don't need to have your source trace in CTF format.
Although, if you
do, it would probably be faster to process (which matters only
if your
trace is large), excluding the one-time conversion cost.
You can create a custom parser for your source trace directly
from a UI
wizard, if it is in text or XML format:
http://archive.eclipse.org/tracecompass/doc/stable/org.eclipse.tracecompass.doc.user/Trace-Compass-Main-Features.html#Custom_Parsers
If your trace is neither text, XML or CTF, you would probably
have to
create your own trace type and parser as a plug-in
extension in
Java code:
http://archive.eclipse.org/tracecompass/doc/org.eclipse.tracecompass.doc.dev/Implementing-a-New-Trace-Type.html#Implementing_a_New_Trace_Type
You can then create your own state system and custom view
based
on it
using data-driven analysis by writing an XML file:
http://archive.eclipse.org/tracecompass/doc/stable/org.eclipse.tracecompass.doc.user/Data-driven-analysis.html#Data_driven_analysis
Marc-Andre has made a presentation at EclipseCon 2015 which
has an
example of what you would need to do:
https://www.eclipsecon.org/na2015/sites/default/files/slides/EclipseConAmerica2015.pdf
(starting at page 25)
Let us know if you need any more help, or if you have
comments or
suggestions after going through the process.
Patrick
On Thu, Jul 23, 2015 at 4:19 AM, Gerlando Falauto
<gerlando.falauto@xxxxxxxxxxx
<mailto:gerlando.falauto@xxxxxxxxxxx>
<mailto:gerlando.falauto@xxxxxxxxxxx
<mailto:gerlando.falauto@xxxxxxxxxxx>>> wrote:
Hi,
I am currently trying to analyze the behavior and
performance of a
system under different configurations and circumstances.
What I have is a series of events (start/stop)
relative to
several
(~200) separate "entities" which I would like to see
plotted in a
timeline: Kind of like the timeline you see on the
"Network" tab of
the Developer Tools under Firefox, where you see how long
each HTTP
request lasts and how it overlaps with the others.
Or like the "Control Flow" view in Tracecompass, except
each row is
*NOT* a linux process/thread, rather an internal
"entity".
So the
data shall not come from LTTNG traces, rather from
external, custom
data.
I suppose I could either:
a) post-process a log file (by writing some translation
script), so
to generate some CTF trace, "faking" processes with my
own
entities, or
b) instrument the code itself so to generate LTTNG-UST
events.
Any ideas how I could achieve this? Just some hint on
where
to start...
Thank you so much!
Gerlando
_______________________________________________
tracecompass-dev mailing list
tracecompass-dev@xxxxxxxxxxx
<mailto:tracecompass-dev@xxxxxxxxxxx>
<mailto:tracecompass-dev@xxxxxxxxxxx
<mailto:tracecompass-dev@xxxxxxxxxxx>>
To change your delivery options, retrieve your
password, or
unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/tracecompass-dev
_______________________________________________
tracecompass-dev mailing list
tracecompass-dev@xxxxxxxxxxx <mailto:tracecompass-dev@xxxxxxxxxxx>
To change your delivery options, retrieve your password, or
unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/tracecompass-dev
