You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is just a very broad issue to collate what we might do if we want Cython to give reproducible output:
It looks like Cython ends up embedding full paths into the source code [ENH] normalize build path #5949. Ideally this'd apply to both c code and comments embedded into the c source.
Do we ever embed a date/time into the output? (I don't think so but could be wrong).
dict in older versions of Python (I'm inclined to treat this as a lost cause)
I think there may be places where the globally defined BuiltinScope, CythonScope or their entries are modified as Cython works. This means that if multiple modules are compiled in parallel then the exact output may depends on the order of parallel execution.
Other parallelization issues? (e.g. there used to be an issue with redirecting error output that was fixed in f372c5a, but it likely isn't the only such issue)
????
There's then the question of how best to test it.
The text was updated successfully, but these errors were encountered:
Describe your issue
This is just a very broad issue to collate what we might do if we want Cython to give reproducible output:
set
s are iterated over and the order is important - e.g. [BUG] Non-deterministic output for local vars with multiple types #5986dict
in older versions of Python (I'm inclined to treat this as a lost cause)BuiltinScope
,CythonScope
or their entries are modified as Cython works. This means that if multiple modules are compiled in parallel then the exact output may depends on the order of parallel execution.There's then the question of how best to test it.
The text was updated successfully, but these errors were encountered: