Glenn Smith
Glenn Smith
Database globals are a bit of a weird place to use, considering they exist outside of the snapshot / undo system. I would recommend trying BinaryView Metadata, which gets saved...
Yeah, in the future we should probably add a big warning notice to the Database family of APIs saying "this is probably not what you expect, try BinaryView metadata instead."
Good to hear. I will still leave this issue open to track the database api changes, which are low priority but easy to fix and should still be done.
Possibly related: https://github.com/Vector35/binaryninja-api/issues/7445
What's the value of `current_function.symbol.raw_name`? The type name may be coming from that, or could be some artifact of the id being empty.
Yeah, looks like the demangler doesn't assign a type id to it properly: ``` >>> (t,n) = demangle_generic(Architecture["aarch64"], "__ZN27AMDRadeonX6000_AmdHdcpProxy12initWithInfoERKNS_12HdcpInitInfoE") >>> t >>> t.parameters [AMDRadeonX6000_AmdHdcpProxy::HdcpInitInfo const& ] >>> t.parameters[0] AMDRadeonX6000_AmdHdcpProxy::HdcpInitInfo const&...
Thanks for the report, looks like this has been causing some pretty nasty mistyped symbols when you only have the demangler for type information. Fixed in ~>=4.3.6467-dev~ >= 4.3.6468
Er, >=4.3.6468, copied the wrong version
Yeah, it's the one going through CI right now, should include commit 74779826
CI took a while cause it actually took two builds before it succeeded :P Looks like this part of the issue is fixed for you, just what remains is what...