I can teach the expresion-coverage-intermediate-variable logic about a bunch of different system functions and what their return types are.

After V3Width all of the system call expressions have proper dtypep()'s, so I'm not understanding the issue is knowing the return types. It's true there's no current way to know what was a system call, you could add a AstNode virtual method to determine that.

Yeah, I'm sure I just didn't look hard enough for the dtype. An AstNode parent would probably be cleaner than my other idea. I can give that a shot.

We can't have a common AstNode parent as some are AstNodeBiOps or otherwise, you need a virtual method.

Oh sorry, I misunderstood. Thanks for the clarification.

I have something ready for this now but it depends on #6589. I'll update once that lands.

@wsnyder I think this one should be good to go now.

How was the list of Ast's getting isSystemFunc determined, e.g. why no AstFOpen?

Grepping around, cross-referencing the LRM and a little bit of Claude. That one should definitely be on the list. I'll do another pass.

I couldn't find a simple list of system functions as opposed to tasks. The lists in 20 and 21 sometimes say a group is just functions or tasks but sometimes it's tasks and functions. I haven't found one yet, but do you know of any place where a complete list of just system functions exists?

All I can suggest is look at the system function list in verilog.y/verilog.l.

Yeah, for better or worse those files also combine tasks and functions. I went through 20 + 21 again and think what I've got now is correct or at least much closer to it.

Thanks, ideally we'd get code coverage on all of these new functions so we have a chance at correctness, but I can appreciate that might be too much work and am willing to waive it.

Taking a look at a quick and dirty test to exercise more of these methods. But I'm not getting expression coverage for:

    if (($time != 0) && foo) bump <= bump + 1;

where foo is a logic. Gonna put this in draft while I figure out if there's something else that needs to be sorted out here.