Peter Schachte
Peter Schachte
The problem is that a fair amount of code is automatically generated in such a way that there is no sensible source location to attach. I don't think we can...
This is hard to fix because of the way type checking runs first, followed by mode checking. Considering only types without looking at modes, we can't distinguish unary negation from...
Yes, exactly. I agree about the source position, but also the code for generating the equality predicate should not try to generate one in the first place when it won't...
Since resources are now actually passed through global variables, that could be used to implement this.
Oops. Fixed.
It's worth reading [Oxidizing OCaml: Locality](https://blog.janestreet.com/oxidizing-ocaml-locality/) for a nice discussion of a similar idea in the context of OCaml. Their approach requires locality (freedom from escape) to be declared, whereas...
An alternative, which is probably more generally useful, would be to implement a declaration of a disjunction of tests which is promised to be true. For example: ``` promise x:int...
Interesting.... I'm not sure why this is happening. I suspect it's because when I create recursive procs to implement loops (in Unbranch) I just make every variable in scope an...
Yes, this is rather fiddly, and as far as I can see, requires processing to a fixed point if you want to be precise. I pushed a commit a few...
Yes, a dependency analysis will be crucial, but I think the reordering also needs to be aware of the different possible specialisations of operations. The reason to do the reordering...