Mihai Budiu

We could add a flag to ddlog to differentiate between these two cases. Frankly, for an incremental engine that case is rather extreme

I have renamed this issue; we should give users the option not to evaluate programs in the initial state.

One reason that this kind of optimizations are not done is that (AFAICT) we actually do not have an IR that represents the dataflow graph before Rust code is generated....

How can you find out in DDlog whether you are in a CLI mode? Perhaps we can have a "system" relation that exposes such attributes. And an ENV relation that...

How about declaring hd as volatile?

That is weird, you would not expect this program to be so large. Maybe you can play more with volatile; perhaps instead of a second structure hd__ you can just...

I frankly do not think this is a problem with our compiler; here are other people complaining about a similar problem: https://forums.whirlpool.net.au/archive/9znlnrk9 The error message is somewhere in iproute.

On the other hand, we are using some hand-crafted bpf headers, which may become incompatible with the kernel headers if they change. This has been a long-standing problem. I understand...

If I understand right, this should be very easy to do as far as the compiler is concerned.

I have pushed a PR against the compiler adding LPM table support. https://github.com/p4lang/p4c/pull/1962 I don't have a recent enough kernel, so perhaps you can help us test?