fulcro app state should contain anything that is part of your current UI, or where you expect your UI to be, also a good caching mechanism, and because it’s normalized it is kept in sync, you can never have duplication of data as long as it’s correctly normalized

but you’re right it doesn’t support arbitrary queries

you could use pathom resolvers to query it in more sophisticated ways, like the example you suggested, BUT it’ll never be as general as a database query engine

From the docs:

The params of these functions can include most of the normal Fulcro load options (such as marker, which defaults to false for state machines), along with these special values:

just use the options from regular load

got it, thanks

no inspect data on workspaces by default, we delegate that to Fulcro Inspect

is there something preventing you from using fulcro inspect? (like using other browser than chrome)

no I do use Fulcro inspect, it's just a lot of clicks to drill down into the data structure to view the contents. I find it easier to just see the data on the card directly. But I realized you can achieve that fairly easy anyway by just adding (dom/pre (with-out-str (cljs.pprint/pprint props))) to the card

workspace cards are also extensible, so you can create your own toolbar and give custom interactions from it (like logging to console)

another tip, if you use cmd+click to expand on Inspect, it does a recursive expansion, so avoid having to click too much

oh that's really helpful! thanks

what's the non-mac keyboard equivalent? EDIT: nvm found it, it's alt-click

Thanks, I had already found tree->db. "If only I could have this in reverse". Back to drinking coffee 🙂

not directly, but you can use parameters to accomplish that, something like: [{(:my-collection {:limit 500}) [:sub :query]}]

Have never seen this problem myself. I would guess you have some kind of conflicting plugin in chrome. Disable all other extensions and see if it works…

Hm. Inspecting the Fulcro Inspect Tab gives me this error: Uncaught DOMException: Failed to read the 'localStorage' property from 'Window': Access is denied for this document. (main.js:3356)

And yes, read/write website data is allowed

And:

Ignoring unauthorized client request from null

_registerExtension @ shell.js:7789

@tony.kay Same error

@wilkerlucio looks like it is accessing the local storage, but not sure why

is this an unmodified output of a template, or have you messed with versions and deps yet?

well, we do use a lot of local storage, what seems weird is why is that getting a denied error, local storage should be accessible

I remember there is some config in chrome that might disallow that, let me check again

@U4VT24ZM3 can you try this and see if works? https://www.chromium.org/for-testers/bug-reporting-guidelines/uncaught-securityerror-failed-to-read-the-localstorage-property-from-window-access-is-denied-for-this-document

Ah! It works if I allow "third-party cookies"

Yes it works. 😄

something to add to the docs

but we can surely give the user a better error message, @U4VT24ZM3 can you please file an issue on fulcro inspect so I can track it later?

Sure

@tony.kay @wilkerlucio There should be a fix, where it is not required to allow them. 😕

@U4VT24ZM3 I think we can wrap the calls to localStorage and detect if it fails, the extension can still work without it, but you lose some features like remembering the watchers and snapshots, so I'm thinking we can add a banner or something let the user knows that's happening

That would be great. It was a chore to learn fulcro without having a clear view on whats going on.

There is an issue, but no solution: https://github.com/fulcrologic/fulcro-inspect/issues/56

Might be related but not entirely the same. I have been using tempids for days without problems

Also fulcro-inspect won't work again

Without using incognito

I can do tempid resolution in my incognito browser with no problems

Closing all open chrome instances got it back to normal

the component for the workspaces must have query and ident, otherwise you can't change it anyway, if you wanna just render you can use a react-card

thats the second time someone get bitten, this needs proper detection as well

So I made it worse by reversing everything to the bare minimum... ident is not necessary,

{:query []} ist enough

I think @wilkerlucio is saying you must have an ident on a component used like that with workspaces

ident is necessary

unless you use ::f.portal/wrap-root false but that’s a different matter

i’m guessing core async channels

what are you using them for?

:shared is meant for immutable data, i don’t think you want to put channels in there

For example, onChange of a text box, I will insert into a channel the new text, it will "do a long process" then mutate the state with the result

Using a async channel I can do that with a slide channel and discard the "excess" of the updates

if we’re talking about a javascript runtime here, that long process will block the main thread regardless of where it happens

you should do that logic in a function that gets called inside the action part of a mutation, then save the result to app state

you could also debounce the onChange function if it’s too much chatter, i’ve done that on color pickers

i don’t think fulcro is meant to be used like the way you’re suggesting

you’re welcome to build your software in whatever manner pleases you, and it’ll probably work for most cases

just know that what you’re doing is not the recommended architecture and you may run into issues later because of it

@U2J4FRT2T Would having an atom in you're utils namespace and just putting those there not work for you're use-case ?

@currentoor @U2J4FRT2T actually shared can hold anything, we had this discussion before, and I for example always use shared for mutable stuff I wanna to be hold with my app

@wilkerlucio but the same can be said about app state too, no?

The only issue I foresee with either approach is things like support viewer and history tracking wont work

its different, because app state we try to serialize in many cases, but not shared

Ah I see. Yes you are right, I take back what I said.

Nice. I will put a lot of non-serial things out from db into shared

non-serialized