Open rrnewton opened 8 years ago
Use of Racket is implied by the ()
, you don't have to have it on the slides. :)
Syntax highlighting would be nice.
6 minutes in, we're finishing up background
More visuals, less text.
Re syntax highlighting, color distinction between bound variables and quoted data would be nice.
That bulleted list of the type inferencer, program generator, CL reducer, could make sweet visuals
... or just distinguishing quoted data, for that matter.
I'm confused about the organization of the introduction. You talk about miniKanren, then microKanren, then back to miniKaren on the next (book cover) slide -- why move back and forth, and more importantly, what is the goal of your talk?
I feel like this entire constraint discussion could have happened before introducing microkanren.
For syntax highlighting, you could follow the typesetting of the Reasoned Schemer book?
Yes, you go through why miniKanren hurts for constraints, and then you could lead into microkanren and talk about adding constraints to it!
Is the headline "tiny implementation of logic programming now features tiny implementation of constraints"? (Because tiny is teachable) Then put the desiderata slide as FIRST slide.
:fireworks: ^^^^^^ what ken said
Agree 100% with @ccshan and @cgswords.
Never use a fixed-width font for text.
Okay we're getting into the good things, and we're already at 10 minutes (the Constraint Recipe slide). That intro should tighten up.
How did the font get so small?
Maybe one way to demonstrate the pedagogical success of microKanren is to jump right into explaining it (without going through miniKanren - just say that it is too big).
@ccshan "I'm not even going to bother explaining miniKanren; it's just too big. I'm going to show you some microKanren, and you'll have no problem understanding it" ?
The syntax-rule could get a nice picture to explain it.
The same for your call/fresh. A little logic puzzle diagram could be cute? Or some sort of reasoning as we move through the constraints there.
The claim of "extensible" makes me want a list of extensions that have been made. I assume: booleans (subsuming SAT solver), numbers (subsuming some kind of SMT), etc.
Complicated macros on your slide seem unnecessary. Also, why does S0
need to be unhygenic?
Put "120 lines" claim on the first slide too.
Using the System, example 1, hits 20 minutes
Write a pretty-printer, use it, sell it as a front-end?
I feel like match
could be used usefully somehow in the type inference example.
@samth are you proposing generating match expanders for mK constraints? Because that's a neat idea.
Time: 22:22
Love the 10-foot font