Jack Huey

@jackh726

Joined on Nov 8, 2019

  • "Ideal" world As PRs are opened, they get reviewed quickly enough that: Other work is not blocked Authors do not "move on"If the author is "new" and trying to learn, this may be "quick", so they don't mentally lose track of what they did If the author is "experienced" (member of the project or regular contributor), this may be "less quick" PRs do not sit so long that their status becomes unclear It's easy for PRs to bitrot, where they now are both sort of "waiting on review" and "waiting on author" (to rebase)
     Like  Bookmark
  • 2023-01-23 POPL meeting with egg crate people Option 1: use egg to replace ena, essentially, growing the ability to have T = u32 and other such things. Have to modify unification procedure accordingly and have some sort of "saturate" step when you encounter projection types. Only have to saturate a projection type once. Saturate(projection) = Solve NormalizesTo(projection, ?T) and for each solution ?T add projection = ?T to the egraph Option 2: some sort of meta-trick-desugaring-thing that Niko didn't 100% understand, basically converting to a form of datalog. Lattice
     Like  Bookmark