Repro and Fix for issue #2529

[crusso]

No description provided.

[dfinity-ci]

This PR does not affect the produced WebAssembly code.

[crusso]

@nomeata PTAL moved the repro to test/mo-idl, it's proper home

[crusso]

So the bug is that the Stamp map, used to distinguish instantiation of the same constructor for monomorphisation, was being shared between constructors with the same name but different stamp (and possibly completely different defs). I've now added another outer level of indexing (__k_n) using k to distinguish cons with the same name but different stamp. I suppress k and n when either is 1.

Which raises another question, why do we number form 1, not 0, so the first renaming is always _2 (not _1)? Was there a rationale for this?

In another branch, I'll experiment with normalizing the type to avoid all these weird indirections.

[chenyan-dfinity]

Thanks for fix!

Which raises another question, why do we number form 1, not 0, so the first renaming is always _2 (not _1)? Was there a rationale for this?

Yeah, it's an arbitrary choice, 0 is probably better.

[crusso]

Thanks for fix!

Which raises another question, why do we number form 1, not 0, so the first renaming is always _2 (not _1)? Was there a rationale for this?

Yeah, it's an arbitrary choice, 0 is probably better.

Ok, I'll renumber and also normalize defs.

[crusso]

@chenyan-dfinity PTAL

[chenyan-dfinity]

LGTM