Andy Fingerhut

A regular PR should be fine. Note: There are multiple ways to implement such a change, e.g. (a) Changing BMv2 so that it reads a new format of BMv2 JSON...

> Usually, there are 2 approaches that programming languages with parametric polymorphism (type parameters) take: > > * Type check functions separately, with the generic types. This usually requires some...

@antoninbas I have started a PR on p4c to enable the p4c bmv2 backend to create indirect tables with a `default_entry` key, here: https://github.com/p4lang/p4c/pull/5104 In testing it, I found that...

In looking at some of the behaviral-model code, I am suspecting that perhaps its current indirect table implementations _do not_ implement default_action behavior. By default_action behavior, I mean: + do...

And in hacking a bit further, _if_ we wanted p4c, behavioral-model, and the P4Runtime API to all support indirect tables with a source-code-configurable default_action, and a runtime configurable default_action, it...

> This is because controls cannot be nested in controls (or functions) and functions cannot be nested in functions, but functions CAN be nested in controls. So the code to...

This requires more double-checking to see whether it is actually true, but _perhaps_ this statement is true: + `base_default_next` is _only_ used by BMv2 for tables that have no `default_entry`...

I have today learned the following by compiling this p4c test program: https://github.com/p4lang/p4c/blob/main/testdata/p4_16_samples/action_profile-bmv2.p4 and running a simple PTF test that sends a packet through, that always misses on both tables,...

Thanks for that info, Davide. Bison outputs source code files with a GPL v2 license header in them, but also a paragraph beginning "As a special exception ...", like found...

This test program is old enough that it is written in an older version of the P4 language called P4_14, whereas the current version is P4_16. They are quite different...