Indexed type for WebAssembly isn’t novel. I guess. I don’t even have a joke about that one.
Conversation
Will you please compare to CertiCoq? Oh sure, you see, they use ANF because I cornered Zoe and Andrew at POPL and tried to convince them that ANF was better than CPS, based on the early result of this paper you're reviewing.
2
1
20
Now they use ANF because I managed to convince Zoe (not Andrew though) to look into it. They've been published cause y'all love a massive Coq proof but not the fundamental research that enables it.
3
1
24
Just wanna say, thanks for all the effort on the ANF front, wrt. compiling dependent types. You and have me pretty curious about it and I'm thankful for your efforts, in spite of the setbacks. Really hoping to see this paper eventually!
1
6
:) You can read Paulette's thesis. We might go ahead and post an arxiv version. This submission isn't dead in the water yet.. all B's isn't the worst set of reviews I've come back from.
5
8
Will do! I'm wanting to make an ANF compiler for my dependently typed experiments, so I'll go to Paulette's first – I was originally looking at your thesis. I'm pretty slow at understanding this stuff and I find implementing it from a nice description is the best way I learn. 😅
2
1
You should look at Paulette's; she's scaled it up to handle if, and recursion via eliminators. Even have a sketch of how to handle match and fix, but preserving the guard condition is on you.
2
3
Dumb question, could you preserve guard conditions by elaborating to eliminators (ala Goguen et al) before compiling?
1
Yeah, but I would really worry about performance. The eliminators you'd generate could be absurd, and I FEEL like it would be easier to compile and optimize match + fix.
I think disagrees about having match+fix in the core and I need to investigate.
2
2
I hope I'm wrong because both sized types and the guard condition are a nightmare, and eliminators are so easy.
1
3
I really hope I don't need match in my core 😔
I was hoping to elaborate pattern matching to eliminators, but I keep hearing claims that it's bad idea for performance, and then others claiming it's not bad, if you do it right. I don't really know either way, but aesthetically eliminators seem nicer?
2
1
Show replies