[news.eclipse.technology.memory-analyzer] Re: New versions of DTFJ adapt

[andrew_johnson@xxxxxxxxxx (Andrew Johnson)]

Batch mode does work with IBM dumps. See this:

http://wiki.eclipse.org/index.php/MemoryAnalyzer/FAQ#How_to_run_on_64bit_VM_while_the_native_SWT_are_32bit


On platforms other than AIX 64-bit you can also use the MemoryAnalyzer 
executable. However, you need to do something special however to get the 
DTFJ Adapter loaded and running.

Modify configuration/config.ini

and change
osgi.bundles=org.eclipse.equinox.common@2:start,org.eclipse.update.configurator@3:start,org.eclipse.core.runtime@start
to
osgi.bundles=org.eclipse.mat.dtfj@4:start,org.eclipse.equinox.common@2:start,org.eclipse.update.configurator@3:start,org.eclipse.core.runtime@start

The reason is because the DTFJ adapter dynamically declares to MAT which 
DTFJ files it supports depending on which DTFJ implementations are 
available in the MAT application. However for this code to run the DTFJ 
adapter must get started. In GUI mode there is an extension point I can 
use to get the DTFJ adapter started. I haven't found a corresponding 
extension point in batch mode. This modification to the osgi.bundles 
statement ensures that the DTFJ adapter is started early and can then 
query the DTFJ implementation and notify MAT which files it supports.

Once the adapter is part of the MAT distribution we might be able to do 
something better.