Nate Foster

Results 224 comments of Nate Foster

I ported the build system to Dune and updated for modern versions of the Jane Street OCaml libraries. Can you try again?

If you try ``` opam pin add yates https://github.com/cornell-netlab/yates.git ``` and then follow the prompts, it should work. Please let me know if you have issues with dependencies. It works...

Sorry, I lost track of this. Would you be able to try with a more up-to-date base OCaml installation? To do this, you would issue the command ``` ocaml switch...

Hmm. This repository has many different routing schemes. @prvnkumar do you recall how to invoke the one that corresponds to Smore? Is it `MwMcf`? https://github.com/cornell-netlab/yates/blob/master/controller/Yates_Controller.ml#L23

True, but `simple_switch_CLI` rejects additions into `t` with action `a`... I think `p4c` should really preserve uniqueness of named actions during compilation (rather than fixing the control plane to disambiguate).

Why are spec changes needed? When the compiler duplicates `a` to get `a_2`, I think it should drop the `name` annotation.

I see, I see. So `simple_switch_CLI` is really the thing that's broken and I should file an issue on Bmv2?

But they are? https://github.com/p4lang/p4-spec/blob/main/p4-16/spec/P4-16-spec.mdk#L7750 The compiler should be fixed though. On Fri, Jul 5, 2024 at 4:41 AM Andy Fingerhut ***@***.***> wrote: > My evidence: > > - They are...

I personally think this makes sense as a future enhancement to the spec and compiler.

Good point. I personally think it would be okay to *not* track local vs. non-local compile-time known value-ness in the syntax. Of course, internally, an implementation of P4's type system...