Descriptive Problem Summary:
515.1609 release crashes when calling string style dlls using call_ext
ID:2880096
Jul 18 2023, 10:52 pm
|
|||||||||||||
Resolved
| |||||||||||||
Are you able to narrow this down at all, preferably with a test DLL project I can compile?
My old test DLL project (I forget who wrote the project for the DLL it uses) is not crashing with this when I pass strings to it, although it doesn't do anything with those strings. Now I did change up the way the memory for the strings is allocated, and that's really the only thing I can think of that could be causing issues. I can try switching back to the old way, but I can't confirm if that fixes the issue if I can't replicate it. |
https://1drv.ms/u/s!AjVBVtsdDUrrg8kja0anvvpGqg7Lmw?e=eQSeul [drathek@drathek-ms7882 bin]$ ./DreamMaker ../../Tester/Tester (I did leave an accidental duplicate define for RUSTG_CALL in world.dm, but even if you remove the extra one the problem is still replicated) I do not reproduce the problem on windows with this tester; also I do not reproduce the problem using a newer version of rust_g. The version provided in this tester should be 1.2.0 of rust_g. Apologies if this is actually just a weird rust_g issue with the version we happened to be using at the time, but MSO seemed fairly certain this was a byond bug. |
This also may be related, but for above with my current linux installation (Manjaro on 515.102-1), if I LDD that .so file it has two missing links:
[drathek@drathek-ms7882 bin]$ ldd ../../Tester/librust_g.so Whereas the latest version of the rust_g library which works w/ the tester has the links found: [drathek@drathek-ms7882 bin]$ ldd ../../librust_g.so However, it doesn't make sense why going from closed beta 1609 to non-closed beta 1609 with I assume the same rust_g binary broke for us, and then returning back to closed beta worked again. So actually I may also be failing to replicate the issue in this little tester that uses rust_g, and if so I'm not really sure what else might be related. |
We were able to swap back to the closed beta and operate normally again.
The line number faulting is
where WRITE_LOG is
bringing us to
where rustg_log_write is
and RUSTG_CALL is
So the result should be:
IIRC this would be rust_g version 1.2.0. We have not tried again bumping to a newer version (2.1.0) yet.