Sanitizing primitive fields

[sfc-gh-shabrainski]

First of all: Thank you for the tool!

We are currently looking into this tool for heap dump sanitization. From the sanitizer 's code and sanitized heap dumps we understand that the sanitizer currently does not redact field values of primitive type and that there is no configuration to do toggle it. Is this correct? Can you explain if this was intentional/ what problems occur if we would be starting to also redact those fields, e.g. for analyzers like JXRay into which we plan to feed the sanitized heap dumps? Happy to contribute if you don't see major roadblocks

[maxxedev]

Set --sanitize-byte-char-arrays-only=false

See more details added recently in https://github.com/paypal/heap-dump-tool/blob/main/README.md#explanation-of-options

[sfc-gh-shabrainski]

Hi @maxxedev, thanks for pointing it out and for missing out on this option!
So from what I understand from its description and the resulting sanitized heap dump, this option also removes any references to objects, right? I guess this was done for simplicity? From our perspective a heap dump would be much more useful if it would sanitize all field values except for references. This would allow analyzers such as JXRay could understand the object relations. In fact when computing heap dumps sanitized with this option JXRay crashes [1]. Do you think that adding an additional option keeping the object reference would be hard to add?

[1]: If this is because of the missing references I am not sure as the stacktrace is redacted and the tool is closed-source but I think it is quite obvious that tools without