ok, I guess this applies: https://github.com/Day8/re-frame/wiki/Using-%5Bsquare-brackets%5D-instead-of-%28parentheses%29

@mikethompson is this crazy? https://gist.github.com/isaksky/2cc3779b52305e992c6150863454a9af

that seems to solve the problem i was talking about

@isak No, not keen on debouncing subscriptions. Just not the right place. Events change things. They do re-computations. They should handle debouncing, etc. Subscriptions are a much more passive set of dominoes. You are trying to create stateful subscriptions which .... is ... bad. > What if the check isn't the problem, but the number of places? Why don't you put the check on everything? If X changes them dispatch Y. Identity check on X, so virtually no overhead. (And then handle debouncing via a debounced dispatch)

You can easily add a interceptor to every event handler Other than that .... do it this way (see point 2) https://github.com/Day8/re-frame/issues/255#issuecomment-274303581

@mikethompson Hm, isn't there more state for the user of the functionality that way, though? Because they need to worry about starting and stopping the :subscription->event with that way. With the function I had, at least the state is just internal.

I see that there is a new reagent with-let function that can help, but then you'd still need to write event handlers that actually spit out the right effects to start and stop

Instead of relying on the :on-dispose logic that happens naturally with subscriptions

So the logic becomes a lot more spread out

I could do the check for every handler, but its not just one value, it is a handful. Wouldn't it also become problematic if that became a standard pattern in the codebase? Potentially a bunch of things that got checked for everything. Seems a shame to not be able to use the normal reaction logic

hi all , wanted to find out if there is a way to effecitvely stream/load parts of my re-frame app as needed , so when i hit the site initially only the code that is needed to load the login page is sent to the browser , once i login the rest can be loaded

@mikethompson is the re-frame lein template kept deliberately behind on some dependencies? I see it was updated a few days ago but lein ancient revealed quite a few older libs and i wondered if there might be compatibility issues that were driving those?

i've never tried it out @rc1140 but you may want to start here - http://swannodette.github.io/2015/02/23/hello-google-closure-modules

@mccraigmccraig thanks , will check it out

i think i remember seeing that post before , but will check it out again

was hoping for some way to do this cleanly within re-frame

assuming the modules stuff works smoothly it looks fairly straightforward to split a re-frame app up, as long as you do it along namespace lines - perhaps with a core.cljs file for each module which makes sure all the subs, handlers and views for that module get required ?

What is a good way to log just the name of the events that are dispatched by re-frame in the console?

For debugging that is

@borkdude that isn't re-frisk?

I didn’t check that out yet

i think you will find it exactly what you are looking for, unless you absolutely want them in the browser console rather than a pop-up window

it also has an inspector for the app-db

essential in my book

I’m going to try it

Should it come up with a separate window?

I’m not seeing anything 🙂

@sandbags I’m trying this now: https://github.com/flexsurfer/re-frisk#web-applications-with-re-frame

Ah I see in the demo that there is something in the bottom right corner

Didn’t notice it in my own app 🙂

It works very good, thanks @sandbags

Hi, just to say that I just finished my first webapp with re-frame — took me a week, with lot of change coming from the end customer and all the while learning the framework : it was fun, fast, and a pleasure to debug. If only everything could be so nice… next web app challenge : using re-com, maybe some React component. Let’s hope my joy will not be crushed !

Can I make the left panel where the event names show wider in re-frisk?

I still can’t read many of them 😉

@borkdude motor4ik is your man for that

stupid autocomplete

anyone hit "Figwheel Client Version 0.5.7 is not equal to Figwheel Sidecar Version 0.5.9. Shutting down Websocket Connection!" ?

i'm looking in the deps and i cannot see where the version 0.5.7 client is getting roped in from

sigh... despite being a new app this is something cached to do with using localhost:3449

whomever it was that said cache invalidation was one of the two hardest problems in CS was not likely wrong

Anyone use run! from reagent.ratom in a re-frame app? http://blog.ducky.io/reagent-docs/0.6.0-alpha2/reagent.ratom.html#var-run.21

@isak when would you imagine needing to?

@sandbags e.g., trigger expensive computation when something changes

have you read about how re-frame subscriptions work?

i'm not seeing the need to do this manually

yea, but re-frame subscriptions are not good for expensive computations that may get triggered frequently

well, they're going to get triggered as they are de-referenced, right?

here is a workaround i did yesterday: https://gist.github.com/isaksky/2cc3779b52305e992c6150863454a9af . But I was thinking about run!then a debounced dispatch as an alternative

IIRC in the docs it talks about how to avoid computing unnecessary subscriptions

i.e. you don't have to re-compute everything every time db changes

yea, but lets say a user can trigger 10 things in quick succession that requires recomputing an expensive dataset - how would subscriptions as-is be a good fit? only 1 recalculation is required

otherwise i guess my instinct would be to use an event, or i don't understand your problem

well that sounds like you need to debounce the event

not manually dispatch the expensive calculation

if I debounce the event, then the UI does not update, which is not good

you might need to separate your event processing into a sequence of events

it is nice if the UI can update, indicate that it is "thinking", then do the thinking

but, really, i am straying beyond my competence in building re-frame apps, i'll let others with more experience advise you

ok. I have a way, just seeing if there is a less "nasty" way

i think there is, but i'll let someone who has implemented it try and explain

ok

Couldn't you make a layer of indirection? Like event-1 updates the db with a :expensive-calc/thinking? and dispatches further to event-2 which is debounced and does the calculation? Or am I misunderstanding something?

@madstap yea that is a good solution, and normally i would do that. The only problem is there are a number of events that can trigger changes to the basis for an expensive calculation, so then I think it makes sense to "react" to those changes, instead of spreading the logic around many handlers, or applying a global interceptor

Anyone know of any re-com applications other than the demo app?

Hey all. Everywhere I use h-box or well boxes in general I get the following warning: Warning: Every element in a seq should have a unique :key:

Could someone help me understand what this is?

I’m usually iterating over vectors to build the view

@petr it's to do with the way react keeps track of elements. When you generate the items in your vector add ^{:key <unique-id>} before each one

e.g. ^{:key "row-15"} and so on

@sandbags thats going to get kinda ugly. Basically I’m iterating over api-responses. Woudl a good time to do it be when adding it to my db atom?

Is there a pattern around this?

worst case scenario use their hash as a unique id

you don't need to add it to the database, just generate it on the hiccup element

react uses the keys to detect changed elements more quickly

Ohhh I see

So I’m using re-com to handle that

I didn’t see a :key attr on any of thecomponents

(h-box :children [^{:key "abc123"}[...] ^{:key "def456"][...] ...])

presumably you're using for or map to create the children

as hiccup elements from your api responses

Ah so it just goes prior to the element I’m about to create

yes

^{} is the sugar for attaching metadata to a clojure object

I haven’t ever seen the ^{} only as ^:dynamic

as an example

So I’m guessing using gensym is not a good idea here

IIRC ^:dynamic is a short-hand for ^{:dynamic true} but i could be wrong about that

I'm not clear what you mean in terms of gensym

@petr: there is another way to do that if you aren't expecting a lot of changes to the data - I believe you use into, I'll try and find the doc for it. It avoids the need for the unique ID.

i wonder how it does that, i'd be interested to know

@shaun-mahood that’s be great!

@sandbags so it’s everywhere that there is a new element? be it a boarder/title/box/whatever?

@petr no, everywhere that there is a collection, list items, rows, a series of divs etc...

without id's react would have to compare all the elements to see which ones had been added/removed

Ah so basically anywhere the :children key appears

in terms of h-box and v-box yes

@petr: I can't find the docs right now, I'm pretty sure it's in the re-frame docs somewhere - it was one of the docs that crossed over between re-frame and reagent so it may have ended up in a weird place. Either way, the essential thing is that if you call a sequence using for, then you will have to add the unique keys. If you use into [], then you don't have to worry about it. The reasoning behind the keys can be found at https://facebook.github.io/react/docs/lists-and-keys.html#keys - my understanding as it relates to reagent is that if you are moving things around (dragging items in a list, for example) then you should see better performance, but otherwise you can just use into [] and avoid the whole mess.

i can't see how using into instead of for can change the situation

@sandbags: Well, if I could find the relevant docs then that would explain it better 🙂

http://stackoverflow.com/questions/37164091/how-do-i-loop-through-a-subscribed-collection-in-re-frame-and-display-the-data-a/37186230#37186230

Oh perfect, didn't even have to light up the batsignal!

ah so it's about lists vs vectors

for returns a list, (into [] ...) returns a vector

okay that makes sense

so (mapv ...) would also be fine

useful to know

Looks like mapv is an issue based on the bottom of that answer, assuming that hasn't changed

thanks @shaun-mahood & @mikethompson

i tried following the link at the bottom which has only a link to another page that does not appear to refer to mapv

https://github.com/Day8/re-frame/wiki/Using-%5Bsquare-brackets%5D-instead-of-%28parentheses%29#appendix-2