Peter Schachte
Peter Schachte
Yes, on reflection, I think you're right that procs with default arguments can be partially applied. Usually. The key is the type system. Eg, `?z = f(x,y)` can be interpreted...
Agreed, `f(x,y,?z)` should always be equivalent to `f(x,y,*default*,?z)`. Unless the type of `z` is constrained elsewhere to be a higher order proc expecting an input argument whose type matches that...
The proposed syntax could be confused with defining `=` as a foreign proc. I guess that's tolerable, since it's easy enough to write out the definition of `=` as a...
Looks good! Then maybe we should change `for` to be `for all`?
Yes, grammatically (in English), `for each` and `for some` work well. Any ideas for a disjunctive version of `do` loops? Maybe keep `do` as the det version, and use `all`...
Yeah, linguistically, `do some` isn't great, but I think it's better than `partial do`, because `partial` doesn't indicate that we're changing the sense of the treatment of partial calls. I...
Exactly. The loop body would execute until success or failure or break, then in any case the iterate statement(s) would be executed to prepare for the next iteration. On failure,...
Agreed. I think separating flattening from type checking was probably a mistake. It keeps causing problems. It's one of those "it seemed like a good idea at the time" things.
See also #467. One further extension: it would be good if there were a syntax for specifying some argument translation, as well. Eg, the Wybe proc should take a string...
Nice, that seems pretty intuitive. Let's go with that.