[Frontend] calling qjit(keep_intermediate=True) multiple times

[erick-xanadu]

Context: Having a single workspace for a single python function to be compiled could lead to an error when the same function was compiled multiple times or another function with the same name was compiled. It also wasn't entirely clear what the behaviour should be when using keep_intermediate and a function was recompiled.

Description of the Change: Now when two functions of the request the same name, a new directory will be created appending _$NUM.

Benefits: No issues with compiling functions with the same name.

Possible Drawbacks:

Related GitHub Issues: closes #291

[sc-47499]
[sc-46567]

[codecov]

Codecov Report

All modified lines are covered by tests ✅

Comparison is base (2280c80) 99.51% compared to head (fff71e9) 99.52%.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #306   +/-   ##
=======================================
  Coverage   99.51%   99.52%           
=======================================
  Files          40       41    +1     
  Lines        7270     7322   +52     
  Branches      423      428    +5     
=======================================
+ Hits         7235     7287   +52     
  Misses         17       17           
  Partials       18       18           

Files	Coverage Δ
frontend/catalyst/compilation_pipelines.py	100.00% <100.00%> (ø)
frontend/catalyst/compiler.py	100.00% <100.00%> (ø)
frontend/catalyst/utils/filesystem.py	100.00% <100.00%> (ø)

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[dime10]

Maybe that's a silly question, but why wouldn't the compiler driver not just overwrite any existing files if the working directory already exists? I believe that was the previous behaviour in Python.

[erick-xanadu]

Not a silly question! I'm not sure, let me dig into it.

[dime10]

Also a good solution!

I wonder what the trade-off is for handling this in the frontend is though, if it is an issue in the compiler driver. With this design, would it simply be an "API misuse" if the compiler driver was fed a directory that already had files in it?

[dime10]

🦝