Impact: API addition (no impact on existing code)*
IMO it should only be an addition; however, currently 0.U is the same as 0.U(1.W), and it would be a fairly substantial API change that could lead to subtle bugs if we changed it to 0.U(0.W). I'm of the opinion that we should not change that, rather zero-width wires could be created explicitly with 0.U(0.W) and via utilities that currently error out like log2Ceil and log2Floor.
Type of issue: feature request
Related issues: https://github.com/freechipsproject/firrtl/issues/508 https://github.com/freechipsproject/firrtl/pull/530
Impact: API addition (no impact on existing code)*
IMO it should only be an addition; however, currently
0.U
is the same as0.U(1.W)
, and it would be a fairly substantial API change that could lead to subtle bugs if we changed it to0.U(0.W)
. I'm of the opinion that we should not change that, rather zero-width wires could be created explicitly with0.U(0.W)
and via utilities that currently error out likelog2Ceil
andlog2Floor
.Development Phase: request
What is the current behavior?
Zero-width wires don't really work
What is the expected behavior?
Zero-width wires should actually work