Using FFI, I need to use a struct full of booleans members. The struct is shown in this page: https://www.khronos.org/registry/vulkan/specs/1.2-extensions/man/html/VkPhysicalDeviceFeatures.html\|https://www.khronos.org/registry/vulkan/specs/1.2-extensions/man/html/VkPhysicalDeviceFeatures.html

You can see that struct is huge. I want to initialize the struct with only two members being true and rest being false. Using the make-VkPhysicalDeviceFeatures constructor seems like a bad idea. The other aproach is create the struct using malloc and then assign true the members I want to. However I want to program in a functional way.

What do you think is the best one?

https://defn.io/2021/01/19/racket-cs-on-ios/\|https://defn.io/2021/01/19/racket-cs-on-ios/ from @popa.bogdanp

Somewhere you have to know which ones you want to be true; I’d probably go with the large call to make-Vk...

I’m going to break the f button, let’s go

@mflatt At least on Racket 7.8 CS, it seems that, although not documented as allowing it, continuation-mark-set->iterator accepts the same #f shorthand to mean (current-continuation-marks) as does continuation-mark-set-first — and, it seems to enable the same “shortcut” mentioned in the docs for the latter: At least on micro-benchmarks it can be measureably faster.

Would it be safe to rely on this and use it? And should the docs be updated to say that?

If yes: Let me know if it would help for me to submit a PR — or, if it would actually take you less time to make that kind of change yourself. :slightly_smiling_face: Either OK for me, just please let me know.

It looks like CS accidentally allows #f and BC doesn’t, and the same for continuation-mark-set->list*. Probably the right right at this point is to generalize BC and the docs.

I’ll look into it later today.

Thanks! Especially the ->iterator flavor, since the motivation already is to be potentially faster, it would be nice if it supported #f to be potentially even faster-er. :slightly_smiling_face:

A question about contracts. If I have a function with two keyword arguments and I need only one to be provided, both may also be provided but then the value has to be equal. I can get the behavior I want using the following, but I have not been able to do it with a contract, it seems like it should be simple, but I’m clearly missing something. (define (transition #:beg beg #:dur [dur #f] #:end [end #f]) (unless (or dur end) (error "need one of dur end")) (when (and dur end (not (= end (+ beg dur)))) (error "begin + duration != endpoint")) (values beg (or end (+ beg dur)))) The contracts I have tried are as follows. (provide (contract-out [transition (->i (#:beg [beg number?]) (#:dur [dur number?] #:end [end number?]) #:pre (dur end) (or dur end) any)])) #:pre doesn’t trigger if the optional arguments are missing? (provide (contract-out [transition (first-or/c (->i (#:end [end number?] #:beg [beg number?] #:dur [dur number?]) #:pre (beg dur end) (= (+ beg dur) end) any) (->i (#:beg [beg number?] #:dur [dur number?]) any) (->i (#:beg [beg number?] #:end [end number?]) any))])) For this approach I have varied all orders and using both or/c and first-or/c, I know why or/c doesn’t work due to an ambiguous match, but I would expect first-or/c to work (also doesn’t this mean that or/c really should be named xor/c?)

Is there a simple way to do this? Thanks!

In your first contract-out code sample, I think the #:pre pre-condition does run even when the optional arguments for dur and end are missing in a function call. Try printing the values of dur and end in the pre-condition code

(the short story is that a call to (transition #:beg 1 #:end 2) for example would bind the pre-condition’s dur to a special value named the-unsupplied-arg. The predicate unsupplied-arg? returns #t just for that special value—there’s an example using it in the ->i docs near the text “If there are optional arguments” https://docs.racket-lang.org/reference/function-contracts.html?q=-%3Ei#%28form._%28%28lib._racket%2Fcontract%2Fbase..rkt%29._-~3ei%29%29)

to check that at most one of those arguments is missing, you could use this as the pre-condition instead of (or dur end)

(nand (unsupplied-arg? dur) (unsupplied-arg? end))

@blachance works like a charm, many thanks! I think what confused me was how the contract interacted with the default values, which is the say, there is no interaction.

~does someone know how to update the sqlite bundled with racket (in windows)? the version it’s shipped with does not have the new shiny cool stuff™, but i don’t want things to break down either so i want to know if it is safe, (my first guess is that you just have to swap the dll)~

nevermind it seems that swapping the dll does not cause any issues

uhh.. so anyone on racket snapshot 8.0(11/19) have an issue with switch tabs? it just wont work ..

the close button is stuck on being highlighted interestingly

I have the same (windows, 8.0.0.1—2021–01–15 bc), luckly the keybindings still work to navigate around.

Send a bug report to @robby . I know the tabs have been improved lately.