As suggested by @luluyi661, the sandboxing behavior can be frustrating when you're just trying to debug and run some test executions of your program. We should provide a command (accept debug? accept build --here or --go?) that makes it easier to see the results of an build or execution and do stuff with the results. Options include:
Disable sandboxing entirely and run stuff in the source directory. This would require the user to clean up intermediates (or we could help with this).
Immediately change the working directory into the sandbox after command execution so the user can explore. (Or run open or xdg-open to show it in an explorer.)
It should be possible to easily debug each micro-step in the process:
Actually building the code (currently: accept build).
Annotating so approximation is possible (currently: accept log).
Running the program so it produces the expected output file (currently: nothing, really).
Load function in eval.py (currently: accept precise, sort of).
Score function in eval.py (currently: accept approx, sort of).
As suggested by @luluyi661, the sandboxing behavior can be frustrating when you're just trying to debug and run some test executions of your program. We should provide a command (
accept debug?accept build --hereor--go?) that makes it easier to see the results of an build or execution and do stuff with the results. Options include:openorxdg-opento show it in an explorer.)It should be possible to easily debug each micro-step in the process:
accept build).accept log).accept precise, sort of).accept approx, sort of).