Eclipse Community Forums: Memory Analyzer » How to write MAT Custom Reports?

Home » Eclipse Projects » Memory Analyzer » How to write MAT Custom Reports?(Where is the documentation for writing custom reports with the intent of running them using ParseHeapDump.sh)

How to write MAT Custom Reports? [message #1798391]

Fri, 16 November 2018 11:04

Eclipse User

Reading the MemoryAnalyzer FAQ, I learned of the ParseHeapDump.sh script and figured out how to run the canned reports (e.g., the suspects report).

I found javadoc for the report package that implies I could write an xml report definition to run a custom report, but could find no documentation about how to do it. I couldn't even find the xml schema it uses.

Any help would be greatly appreciated.

Thanks

Re: How to write MAT Custom Reports? [message #1798788 is a reply to message #1798391]

Sat, 24 November 2018 16:05

Eclipse User

I've been considering this recently but haven't written any more documentation other than adding some JavaDoc and updating the schema.

An example report suspects.xml might help you too.

A 'section' is part of the report - either a possibly collapsible section or a separate file. A 'template' is a way of adding another report. A 'queryspec' actually runs a command. [You can also test running commands from a command line using the search queries menu.]

Reports can be controlled by a 'param' - there are several ones defined and documented in the Params Javadoc. The latest builds have some updates to the Javadoc.

The params can be used to control generation of tables - as HTML or CSV files, and to omit, sort or filter columns. Some params just control the current section - some also control any inner sections unless overridden.

A param can also be used elsewhere in the report as ${param_name} - in the command name and other param values.

I have been considering how to make reports more flexible. One idea is to allow params to be set on the command line - as:

ParseHeapDump mydump.hprof -myopt=myvalue MyNewReport.xml

where the option is set before the report in the command line. This is still experimental as I'd like some feedback on if it is useful. I used it for the new export HProf report. The ParseHeapDump options have an equals sign - but queries don't, so be careful.

I'm also considering how to deal with multiple snapshots and comparisons. One way is to define the secondary snapshot as an argument to a query, then pass it into the report.
E.g.

    @Argument(advice = Advice.SECONDARY_SNAPSHOT)
    public ISnapshot snapshot2;

delta_histogram -snapshot2 "myseconddump.hprof"

in a report as

<queryspec name="Compare">
  <param name="snapshot2" value="defaultsnapshot.hprof" />
   <command>delta_histogram -snapshot2 ${snapshot2}</command>
</queryspec>

and run as

ParseHeapDump dump1.hprof -snapshot2=myseconddump.hprof compare.xml

That's not very flexible, especially as there are inbuilt compare queries. Perhaps the param name could be employed in a future MAT. Defining it might define the current snapshot for queries. I'm not clear how to prove tables as inputs to the compare query. Possibly as <spec> sections inside a queryspec. They could then be made available to queries with IResultTable/IResultTree arguments.

Do have a go at writing some queries - and let us know what works and what doesn't.

Re: How to write MAT Custom Reports? [message #1800006 is a reply to message #1798788]

Tue, 18 December 2018 14:22

Eclipse User

Thanks for your detailed reply. Currently, the histogram output in csv format with a limit of 1000 was all that I needed. Hence, tweaking the overview.xml report was all that I needed.

I will bookmark this thread just in case I have a tougher memory issue and need to do snapshot comparisons.

Thanks again.

Previous Topic:	Meaning of Overview->Details->Size
Next Topic:	cannot load hprof file with heap dump error

Goto Forum:

-=] Back to Top [=-

Current Time: Sun Jul 20 14:06:38 EDT 2025

.:: Contact :: Home ::.

Breadcrumbs

Sign up to our Newsletter