Need a test case.

as i said, is hard to find a small test from my designed CPU, because this bug may occur in complex situations.
(actually i am not familiar with the verilator source code, hard to create a test through reverse analysis)
i can provide my project for testing

Without a small test it's likely it will break again.

I am not sure if it is correct. I think it is possible that two types that should be considered the same are different AST objects. Generally, types should be compared with similarDType method. If it doesn't work, there is probably a bug.

I am not sure if it is correct. I think it is possible that two types that should be considered the same are different AST objects. Generally, types should be compared with similarDType method. If it doesn't work, there is probably a bug.

sure, i will try it
but so far i haven't found a small test

i think i find the real problem
the wavedump's value is not real value
a simple case:

assign a = 0;
assign a = 1;

the wavedump's value is 0, but the real value is 1
this pr will make wavedump's value equal real value and final both equal 0

this made me misjudge
i think need another way to fix it
this pr can close