Closed ChrisJefferson closed 1 year ago
good find, thanks. does your branch do any better @Maurice-Byrne ?
It doesn't crash, but the error message is still a bit cryptic.
Yes, we need a check for record/variant field names. I'll add this.
Not claiming this is the best error message, especially since it doesn't include line numbers, but it's an improvement. Maurice can add the line numbers :)
In the following spec I use blocks as a variable name and record name. The constraint is considering
blocks
as the record member.This is (possibly) two bugs:
1) Don't allow name collisions if these aren't independent namespaces. 2) Maybe a slighter nicer error is someone tries to quantify over a record name (which is, I think, just nonsense)
This issue came up because I thought I'd be "clever" and store the neighbourhood with the same name as the variable it is a neighbourhood for.