I presume #racket-school is deprecated in favor of this channel?

Looks like it

In Matthias’ presentation, his walkthrough of macro expansion required macro.rkt and table.rkt, are there standard repositories for these files?

right, where are we eating?

https://docs.google.com/forms/d/e/1FAIpQLSczb6r1HmNJ5gthE8BguYiseZ_IONH09cY5hVhrZMG2gmlwiA/viewform

the survey for today

https://forms.gle/y6QPgnwsXiJv1reJA

If anyone’s interested in talking about API design, enforcement, and/or dependency versioning please come find me - I’m in the front of the room in purple.

we got some interested in VR/visualization

maybe we can be with the gamedevs?

Here’s the Monday survey URL: https://forms.gle/y6QPgnwsXiJv1reJA

Has anyone here used Rosette for any research or personal projects?

Is there any spot Schemers gather?

I have to work out some thoughts on problem 7 (¿syntax-foldr?) so I’m probably just going to eat at the heritage center… but I might go for a drink later if anyone is interested.

@me1150 I’m not at the summer school, but I use Rosette for my research (Emina is my advisor).

what’s syntax-foldr?

I don’t know yet! But probably a horrible thing to write to solve this problem (essentially monad threading)

Ah, you are solving monad-do in https://school.racket-lang.org/2019/plan/mon-aft-lab.html?

aye

Rosette is super cool, maybe we can put together a dinner group tomorrow on SAT/SMT solvers

@zenspider Are you in Heritage + want to work on #7? How far are you?

I’m just flailing right now. I’m in my living room (220). Happy to have you here or to meet you in the cafeteria or one of those study areas on 1 of heritage

wait. not 220… 213?

Down in a sec

Heading over

where is cafe?

Heritage. Second floor. Follow the noise

@sorawee We figured it out, can post some code if you’re interested :slightly_smiling_face:

Nice! :slightly_smiling_face:

@sorawee How the heck does Rosette do combinatorial problems (like SAT or synthesis) quickly?

If it’s black magic you can be honest

So, Rosette itself doesn’t “solve” anything. It simply (efficiently) generates constraints and sends queries to off-the-shelf SMT solver.

Ah gotcha! I’ll have to play around more. I’m an ML person so that recent paper from Emina et al on synthesis CUDA kernels is super interesting.

Will solutions to the lab exercises be posted somewhere?

Here’s my solution to Exercise 7

(define-syntax-parser monad-do
  [(_ return bind [x #:<- e] . rst)
   #'(bind e (λ (x) (monad-do return bind . rst)))]
  [(_ return bind [#:ret e]) #'(return e)])

(not optimized for error messages)

I wish i’d have known about Rosette last year; I’d be very interested in trying a few things out in it. There are quite a few problems in computer algebra where a nicely interfaced constraint solver would go a long way

Same :slightly_smiling_face:

Ugh, sorry wrong format:

(define-syntax (monad-do stx) (syntax-parse stx [(_ return bind [#:ret ret-form] rest ...) #'(return ret-form)] [(_ return bind [x #:<- mx] rest ...) #'(bind mx (λ (x) (monad-do return bind rest ...)))] ) )

So define-syntax-parser is syntactic sugar for the nested define form: (define-syntax...(syntax parse...))

https://github.com/disconcision/racketskool/tree/master/lab1 no claims are made of either correctness or elegance. please do not use as a basis for medical software

Right

define-syntax-parser is usually enough if you want to define a macro (as oppose to doing something weird with syntax object). The only thing that seems to be losing is the whole syntax object itself, but that is recoverable using this-syntax.

I’m not sure what you mean by dispatch on an or* pattern. You can tell which variant was matched if an attribute is bound differently, by looking at that attribute. But that only works if you define an attribute for that.

For instance you could have the a-clauses class define the attribute accepting? = #true and the r-clauses class define the attribute accepting? = #false, and then dispatch on the value of (attribute clauses.accepting?). Does that cover what you meant?