Profiling and name mangling

[William Harvey]

Dear All,

I'm doing some performance profiling on a large Scala project using
jvisualvm. Most of the time I can follow what is going on, but there
are certain times when there is a hot spot in an anonymous function or
some exotic location which looks like a cross between ancient Rome and
a bank vault (lots of Roman numerals and dollar signs), and I'm not
100% sure how to un-mangle the name of the hot spot and map it back
into a line number in my source code. Sometimes jvisualvm gives me a
line number - for example in the Thread Dump view - but in profiling
and sampling results I don't get one.

Is there a guide somewhere that can help me un-mangle method names so
that I can better track down the hot spots in my code? In general, it
would be helpful to know how the name mangling works for a variety of
purposes.

I'm sure people who do scala-java interop on a regular basis have this
all figured out, but I haven't dabbled with this yet.

Thanks!

William Harvey
http://www.cse.ohio-state.edu/~harveywi

[Daniel Sobral]

This should help a bit:
http://stackoverflow.com/questions/9350528/how-to-work-with-javap-for-scala-java-interop

--
Daniel C. Sobral

I travel to the future all the time.