Brian Kelley
Brian Kelley
That looks suspiciously related, however with that PR the multithreaded tests I put in certainly crash so I'm confused why reverting it works.
Are you sure it's not from the new multithreaded test itself? There could be other issues lurking besides what the mutex is preventing. Try reverting just the SmilesWrite mutex and...
I'm 99% that this isn't the mutex, it's the test.
I think at least one of the problems is here: ``` ==216296== at 0x4F7518E: size (stl_vector.h:919) ==216296== by 0x4F7518E: operator= (Dict.h:107) ``` We are copying properties that may be being...
I think the only way to Make MolToSmiles (and RDKit in general) really thread safe is to assume that properties are really const too when the molecule is const.
One solution I hate is to have a mutex on the molecule itself.
Molecules output by RunReactants aren't sanitize (this is for a few reasons) to resolve this, you need to run SanitizeMol on the molecule. Note that this can throw an exception...
It is documented in the rdkit manual in the Chemical Reactions section here: https://www.rdkit.org/docs/GettingStartedInPython.html
@greglandrum feels a bit overkill for two files, but if we decide to do this at configure time, should they be in RDGeneral or the SimPicker?
It's a good question. I'll try a version with External and see what it looks like.