New blog post: Scope Sets as a Piñata for Name Resolution.
typesanitizer.com/blog/scope-set
I describe what scope sets are about, why I think they're cool, how you could use them to implement a name resolution library, and more.
4
10
21
Conversation
Really cool post! I'm wondering if this could be used for dependently typed languages somehow? I'm guessing that Would need to figure out a good evaluation and conversion checking algorithm though… 🤔
1
A big thing I want to be able to do is be more flexible about permutations on application order and field order in dependent function and record types respectively. Wondering if this could help?
2
Would also be interested in how fast this could be made - I would assume that using high level hash sets with strings would be rather slow - would it be possible to use bit sets and stuff to make this more speedy. 🤔
Replying to
For perf, there are many options. Intern identifiers. Scopes are unique integers. Intern scope sets. Cache lookup results. Cache subset check results. There's also the question of multithreading. I don't have a good sense of where the bottlenecks would be.
1