One case of max iteration count being exceeded.

In my project I am encountering the max iteration count problem.
It is working with a list of about 20 modules. After two iterations of analysis, all the modules get deferred, and four of them report making some progress.
The problem is that these modules are making the same progress every time. It is adding an entry to the module's symbol table. However in the next iteration, the symbol is missing from the symbol table and gets added again, resulting in a report of progress being made.
The semanal code is too complicated for me follow. It seems that somehow, the updated symbol table for the module is getting lost and the module state remains at its state before the iteration pass.

I hope this can give you a clue to what to look for in the semanal code that would account for the symbol tables not getting carried over to the next iteration.

I did manage to do some debugging and track down these progress events. One example is in a module:

class InstrValCol(Col[ColKey, ValType], ColAttrsBase, metaclass=CacheMeta):
	...
	class LookupResultTup(NamedTuple):
		status : RowKeyStatus
		value : Optional[ValType] = None

		def __repr__(self):
			repr = self.status.name
			if self.value is not None: repr += f' {self.value}'
			return repr

It is the name __repr__ which is being added to the symbol table every time.

Start by tracing the semanal iteration code around the reported progress events, using the InstrValCol example and its repr symbol as the reproduction case. Follow the module state and symbol table across iteration passes; done means the updated symbol table is retained and the same symbol is not reported as new progress on every pass.