Need ability to fail jvm start on bad vm_args

[clearbluejar]

Looking for flexibility to pass custom args to jpype.start. Well, not necessarily all args, but at least ignoreUnrecognized.

https://github.com/dod-cyber-crime-center/pyhidra/blob/718627f8289019cf0433d6a4e9f535f63cd56d3e/pyhidra/launcher.py#L204-L210

When creating my own launch and passing in vm_args, I'd like to be able to set ignoreUnrecognized to False.

This allows me to error out when an arg I pass and expect to work fails...

launcher = pyhidra.HeadlessPyhidraLauncher(verbose)

launcher.add_vmargs(f'-XX:MaxRAMPercentage={max_ram_percent}')
launcher.add_vmargs('-XX:+HeapDumpOnOutOfMemoryError')
launcher.add_vmargs('-XX:+SOME_ERROR_ARG_THAT_MIGHT_FAIL_AND_I_WOULDNT_KNOW')

launcher.start()

Perhaps we could add an optional arg to: https://github.com/dod-cyber-crime-center/pyhidra/blob/718627f8289019cf0433d6a4e9f535f63cd56d3e/pyhidra/launcher.py#L129-L137

def __init__(self, verbose, ignore_unrecognized=True): 

Happy to submit a pull request. Just let me know.

[clearbluejar]

Perhaps there is more to this than I understand.. just saw this: https://github.com/dod-cyber-crime-center/pyhidra/blob/718627f8289019cf0433d6a4e9f535f63cd56d3e/pyhidra/launcher.py#L47-L48

I tested it by setting ignoreUnrecognized changing it, and was able to get the JVM start to fail on an invalid vm_args, but there must be a reason you have the code above.

[dc3-tsd]

Thanks for pointing that out. This option was needed during early development for prior versions of Ghidra, but is no longer used in the versions that Pyhidra supports. A pull request with whatever changes it now needs would be welcome.

[clearbluejar]

Sounds good.

[clearbluejar]

https://github.com/dod-cyber-crime-center/pyhidra/blob/718627f8289019cf0433d6a4e9f535f63cd56d3e/pyhidra/launcher.py#L204-L210

coming back round to this.. do you also need convertStrings=True ?

Seems like that setting forces Java to cast it's strings, but they don't recommend it for new dev. https://github.com/jpype-project/jpype/blob/6869f03fd37184ccfde4db043b2927837e91e072/jpype/_core.py#L147-L152

Does pyhidra / ghidra have known requirements for it, or also a legacy requirement?

[clearbluejar]

answered my question here. looks like convertStrings is required.

[dc3-tsd]

This was deemed necessary because converting strings manually would have become heavily tedious with its use in dragodis. However, if the consensus in the community is to not have convertStrings​ on as default, we could just set it only within the dragodis code, now that we have that PR change you provided. Alternatively, we could support not having convertStrings​ on, but keep it on by default to provide less jarring results to the end user.

If we wanted to make pyhidra work with convertStrings​ off, we believe it would just need to go through and manually convert the Java Strings to Python strings in a few places like this.

[clearbluejar]

I am indifferent to the convertStrings issue, I don't know enough about it. I assume there is some overhead to convert all the strings but it is helpful when accessing strings from python? As the pull request has been accepted. I am closing this issue.