Jeremy Kun

A colleague pointed me to this which might help support the variadic arguments issue: https://jsdw.me/posts/rust-fn-traits/ See the section "Accepting variable arity functions using Trait Families"

All the fixes are in, and now we just need to wait until the upstream changes are integrated into HEIR, probably will happen on Monday.

We're still missing the last upstream PR that will fix the remaining test failure

I think this second pattern will be a bit more complicated, since Op will be a tensor op in one domain and a polynomial op in the other, and it...

I think the right approach to figure out what to do next is: 1. Add a pass that actually introduces ntt/intt ops by replacing polynomial.mul with ntt + tensor mul...

I filed a separate issue for HEaaN.mlir, and we can limit this issue to the basic canonicalization patterns: https://github.com/google/heir/issues/635 I think loop fusion would be a good place to start....

Now that I've got the openfhe e2e tests working, it seems reasonable to migrate to proper rust rules and a rust test harness instead of lit testing.

> Safer-MLIR paper ... Could you link to this paper? I don't recall it.

> I'm sorry about the typo/auto-correct; it is indeed Syfer-MLIR. @AlexanderViand-Intel - the paper moves out n-invocations of FFT and IFFT; this is important and not come for free from...

Unless there are any more specific canonicalization patterns we think we can add for ntt/intt, I think we can close this issue.