Given that verilator has a way to waive issues programatically (e.g. https://wallento.cs.hm.edu/post/20200612-verilator-waivers/), it would be quite nice to integrate those seamlessly. Especially being able to export verilator configurations that are usable externally.
There are a few concerns that would need to be addressed before going ahead and integrating this.
We still need some way to track the existance of waivers. This is fairly easily done by just parsing the verilator config file. But how should this integrate into the existing system? Do they live alongside current waivers? Replace them completely? Be dynamically serialized from waivers?
Multiple warnings of the same type on one line can be clobbered by the way verilator handles waivers. This could potentially hide real bugs.
Orphaned waivers may be harder to manage, as you don't get any context other than the waiver itself.
Given that verilator has a way to waive issues programatically (e.g. https://wallento.cs.hm.edu/post/20200612-verilator-waivers/), it would be quite nice to integrate those seamlessly. Especially being able to export verilator configurations that are usable externally.
There are a few concerns that would need to be addressed before going ahead and integrating this.