semProcAux hasError cover fixupInstantiatedSymbols

[bung87]

Summary

• semProcAux hasError cover fixupInstantiatedSymbols
• Instantiated symbol's AST may be the nkError kind that causen[bodyPos] raise FieldDefect

Details

• fixupInstantiatedSymbols now returns false if instantiated symbol's AST is nkError kind during iteration

---

Notes for Reviewers

[zerbina]

This is not correct, sorry.

There being an error in the instantiation of a generic procedure is not an error with the definition. Besides, the way you use hasError here is wrong: AST must only be wrapped in a wrapper error (wrapError) when there's an nkError somewhere within, but this is not the case here.

In general, please document in the PR message what exactly you're fixing (what is the error and why did it occur), and also provide a test where you have made sure that it reproduces the failure, thanks. I don't want to have to investigate these issues myself each time.

[bung87]

by the words "not an error", here should continue iteration without marking anything ?

[zerbina]

by the words "not an error", here should continue iteration without marking anything ?

Yes. Instantiations with errors are reported when they're inserted into the module AST.

[bung87]

is that a global error? because here n[bodyPos] may cause FieldDefect

[zerbina]

Oh, my bad, I misremembered. instantiateGeneric reports the error (via localReport), appendInstancedGenericRuntimeRoutines doesn't.

[bung87]

okay so here the error check is needed I guess, I also aware even similar kind of errors, they use localReport or globalReport that seems don't have certain rules. that's another thing.

[zerbina]

To be clear, my original review comment still applies.

[bung87]

okay, so this is the another case compiler continue compilation cause subsequent errors, the appendInstancedGenericRuntimeRoutines do internalAssert which should cause compiler abort,
whereas structuredReportHook may just return doNothing