Jessica Clarke

Whilst I agree that the global is ugly, I disagree that your proposal is a better fit for the architectural specification. Architecturally, illegal instruction is just another exception, and so...

> Good call, that does sound better and I guess would let you treat other exceptions in the same way (I'll have to double check the code though). I don't...

That requirement doesn’t exist in older versions of the spec IIRC

20190608 (ratified v1.11) says just: > The PTE format for Sv39 is shown in Figure 4.18. Bits 9–0 have the same meaning as for Sv32. Bits 63–54 are reserved for...

It should not take 15 commits like that to add this extension. Please condense into a sensible history.

> Pull requests should be a single set of related changes with a clean commit history free from merge commits. -- https://github.com/riscv/sail-riscv/blob/master/CONTRIBUTING.md

Please avoid creating 16/32/64 copies of every function. You should in general be able to write one polymorphic function that handles all sizes.

Whilst float_eq is relatively straightforward to get right, actual arithmetic operations are much more complicated. What is your strategy for ensuring correctness?

> Sorry for disturbing. If there are no more comments to address, can we squash merge this PR when free? Then we can continue the rest API implementation of float...

> > Well my comments remain unresolved, and your responses haven't changed my opinion... seems slightly untoward for you to be saying this when there are outstanding comments you've just...