Anthony Donlon

The problem is that currently we have two linking passes in Verilator (V3LinkDot::linkDotPrimary and V3LinkDot::linkDotParamed), where they're before and after the parameterization pass (V3Param), respectively. V3Param resolves generate statements and...

The fix is indeed too simple and lacks of more considerations... Or maybe can create a Cell for each of the ports, like the original author of the feature ever...

Thank you for commenting this! Yes I agree with you said that it may causes more bugs, since the change causes more nodes require relink. But I think such bugs...

> I think I do want us to keep the original module around In previous version there was still chance that the module will be removed ([here](https://github.com/verilator/verilator/blob/e6135981a56782a4650902c66e5c93eb569cd78e/src/V3Param.cpp#L1437)), as long as...

Hi, I just saw that you mentioned V3Dead somewhere else (https://github.com/verilator/verilator/pull/4646#issuecomment-1789752593) so I wonder if you have started coding? I was also digging into it and conceived some immature ideas,...

> I have a new V3Dead all coded, just in debug phase Thank you! Looking forward to the improvements. > which I underestimated! My humble understanding about its complexity is...

> This looks like would help #4391 and similar. I think #4659 is more likely to fix it. If we process modules in the order of source code, the xxx_class#(Cfg0)...

> Unfortunately source code order is not sufficient, and changing to that will break other cases Are they cases containing hierarchical names? So I think actually we do need to...

V3Param doesn't generate correct new name for some parameterized classes. I think I could have a try to fix it...

Done.