[P4-dev] (P4-16) About header_union behavior
jnfoster at cs.cornell.edu
Thu Apr 20 14:14:43 EDT 2017
> (I've recently tested the latest version of the compiler and found that
> extracting or emitting individual header fields is no longer acceptable.
> The compiler demands the parameter for extract() to be header, stack or
> union. I assumed unions were really being implemented into the language, so
> I snooped the project files and found the header_union construct.)
Header unions are still being designed and are tagged as an experimental
feature in the course code. We are hoping to finalize and release the
design and implementation next month.
> Let's say I extract into a. If I then read/write to b.fixedF, can I expect
> it to yield equivalent results as if I were reading/writing to
No. The current design is more like a tagged union.
> Also, will header_union have an isValid() method? (currently they don't
> seem to have one)
Not currently. Is there a strong reason for having one?
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the P4-dev