/prague/ Q&A

Ask /prog/ anything.

Although don't expect an answer or even a good one.

Also, keep it /prog/ related.

>>61

For it to really work it needs to be interactive. Or the analyzer would need to list all inferred types so the programmer could look through it and find the ones that doesn't match what they thought, and fix the problem there.

This is actually what I was thinking about, thank you for articulating it so nicely. Sometimes this can be difficult, for example if the code is actually part of macroexpanded code, but it should still be feasible. Most of the time the static analyzer will be complaining about the lack of annotations rather than pointing out mismatches.

I think the easiest way to get this would be to have the analyzer output the code in a sub language with explicit types, and then feed this into a secondary compiler.

Or just let the compiler accept type hints (like CLs have been doing since forever) and use them whenever it can.