Alan Lawrence
Alan Lawrence
I suggest a Hugr-side `struct RootView(&Hugr)` that implements HugrView - either via `delegate!` or by `AsRef` - and whose constructor checks the root type of the hugr in the same...
The need for `dup` here is critical. If we didn't need to `dup` then we could take `Lazy` as `QUBIT` and `read` thereof as `measure`.... if (given) we do need...
> They are not exactly the same as `Ptr`s, they have no `write` op. Ah, yes, true. The writing is captured in the closure of how to compute the value....
+1 to Arc over explicit lifetimes. I think multithreaded analysis is not a big problem for the time being. Not sure I understand how the Context stuff would *look* in...
> We would also have the option to go for a global Context Isn't Context's `fn get() -> Self` essentially a global Context? I mean, thread-local and with a stack,...
> We already have to deal with illegal structures appearing: there are plenty ill-typed programs which are still expressible in core. So we already need the necessary plumbing to account...
Not sure if you mean loop unrolling or loop *peeling*. * Unrolling is `do C while P;` --> `do {C; if !P break; C} while P;` (normally only done if...
See https://github.com/CQCL/hugr/pull/2290
@ss2165 yes, indeed we can make this non-breaking in the technical API sense (code, using the hugr family of crates, that compiles before, will still compile). Of course we can't...
> > yes, indeed we can make this non-breaking in the technical API sense > > Unless we can upgrade serialised HUGRs with functions inside DFG regions by pulling them...