@alandipert what was that thing you wrote where you could query vectors?

ah, found it https://github.com/alandipert/intension

is there a hoplon way to do http/rest request or this is just clojurescript ?

@ejemba no standard Hoplon way for that, I’m afraid. There’s Castra, but it’s RPC

Can anyone please explain what the js-arguments is used for in here: https://github.com/hoplon/javelin/blob/master/src/javelin/core.cljs#L193?

@dm3 so the prefered way of working with hoplon is with castra

no, it’s just something that already works

oki

@thedavidmeister so I have a few proposed solutions that I am working on, the first is to move your fix to a new attribute :multi-cell which works like and returning the last value, it takes a sequence containing an attribute and cells. The attribute is bound to each cell. I’m calling this the n(1) binding solution. My second proposed solution is to inform elements about their -tpl index. This could enable a tpl-cell which updates whenever the element index changes. The third option is to always call do! whenever an element is added to the dom.

@alandipert ^^ thoughts??

@alandipert also I am receiving weird messages from javelin java.lang.String cannot be cast to java.lang.Throwable at line 27, column 1 in file app/index.cljs the line contains cell? which it cannot find because I was missing the namespace however this was really hard to spot.

@ejemba js-arguments is a function for referring to the JS arguments keyword in a function… which refers to an array-like thing containing all arguments that were passed

@alandipert it's @dm3

@alandipert figured it out - missed that it’s inside another fn

oops, cool

was porting that ns to clj

@flyboarder maybe you could make an issue w/ one or both proposals?

then i can look at it later and bug micha to do the same 😄

@alandipert I see there’s the Clojure Javelin issue open: https://github.com/hoplon/javelin/issues/25 with some moderate activity

there’s also a javelin.core-clj in a working state, as far as I can see

haven’t seen you cljc branch and did pretty much the same thing, only used mutable variables instead of refs

although my propagate is 10x slower than the one in javelin.core-clj so that’s not great

and I agree that the code doesn’t get any nicer…

@dm3 yup

based on my recent experimentation, i don’t think the api in general is a great port to clj

what do you think are the bad points?

in cljs you can use formulas casually for side effects, but in clj i think more rigor is required

so i think you want something that runs in the same order a formula would, but can’t have dependencies

i think the watch API could be enhanced to be this thing

esp. if the watch function is run immediately, the way do-watch does it

i started to make a new clj version a few months ago but it’s on my other computer

and then i switched to using R lol

do you mean a case where a formula does a side effect and is a source for another formula?

correct

hm, I don’t immediately see it as a big problem. You can probably shoot yourself in the foot a few times with that though

yeah, it’s really not a big problem

and what about threading? I was thinking a single-thread-only graph would make sense

yeah i agree there

that’s another thing about side-efffect-only-formulas (“spouts”?)

is that they could run in other threads potentially

since they don’t have dependencies, they could actually block in a transaction and the transaction would still go

yes

add-watch=

altho, a user could just make formulas or watches that put things to do in a queue

pretty much

either you give the add-watch a function which then executes on your thread pool

or you manage the queue yourself

yeah, I’ll try using your javelin.core-clj

i’ll push what i had a few months ago

iirc it was mostly working and was cleaner than core-clj

my thinkpad is under the couch somewhere, i just need to retrieve it

sure, that would be great

@dm3 turns out, i already pushed: https://github.com/alandipert/plumbata

i remember it had bugs. but i don't remember what they were 🙁