I still want to ask yesterday’s question.

@eric.shao What is the problem with this state img ? 🙂

if you're referring to "I want to load one {} or a [] to append to the initial-state’s list.", this one sounds like you should use post-mutation instead of target to implement this custom logic

but your first image shows it as a vector

Is the return from the server a map that contains the data needed for normalization (ident function on RepaymentItem)?

😅 yes, I am not experienced to figure out the problem.

wait..the answer to that must be yes, because it did normalize

or did it?

you’re showing 888 in both images

the 888 from server is normalized.

but your first image already shows it there…so how do you know?

so it is in :REPAYMENT-ITEM/by-idx

and that came from the server for sure?

yes for sure, if I comment the {:fulcro {:started-callback (df/load)} 888 will not show in client.

image1 is load from server, just not “target”

oh, so your earlier question was “which comes first…“, right? So two are in initial state, and then you run a load in started

what is the second image?

image2 is load from server and target(df/append), but the 888 ident is still there (not disappearing) and the 1 and 2 ident are turned into nil.

so this is out of Inspect? A step by step review of what happened?

yes.so I ask which is first. after some console.log I know initialState is first.

right, but you’re only issuing one load?

ok ,give me a minute.

not two, right?

just one load

it is possible you found a bug…could be initial state and load are not playing right together…but darned if I can see why. Try this just to be sure: Put your load inside of a setTimeout to defer it for a bit:

:started-callback (fn [app] (js/setTimeout (fn [] (df/load …)) 500))

ok

It shows slower which is the same as the inspector’s history. First 1 and 2 is showing,then 888 is merged in ,and at the same time the [1 2] turned into nil.

any chance you can post the code to github? I can take a look in a bit.

ok

@tony.kay https://github.com/EricShaoCN/daikuan/blob/try-semantic-ui/src/cards/daikuan/loan.cljs

https://github.com/EricShaoCN/daikuan/blob/try-semantic-ui/src/main/daikuan/ui/loan.cljc

@eric.shao I don’t see a problem with your code. I’ll open an issue and track it down.

@tony.kay Thank you for your patient.

Sorry you hit a bug on such a small attempt 🙂

I’m kind of glad. Does that count a contribute?😅 Maybe I will never contribute a thing because of my poor ability.

Every little bit helps 🙂

@eric.shao workaround…wrap the append to in multiple-targets: (df/multiple-targets (df/append-to ..))

as the target…that works. I think perhaps I never tested the standalone append case, but always assumed it would be combined

@tony.kay yes , it works!😀

so, that’s an easy bug fix as well…thanks for being persistent with your question 🙂

proofreading contribution - there are two instances of "the the" on http://book.fulcrologic.com

also first list item under section 4.2.2 - "Fulcro provides a a built-in.."

are there docs on contributing to the docs?

another issue - not sure if it's as-designed, but if a component's query function returns a list instead of a vector, it won't validate.

not really that big a deal, but it wasn't obvious why. doing (reduce conj xs ys ..) instead of (concat xs ys ..) fixed it

nevermind, looking at the spec i see it's using the list/vector type to determine template/method

@chrisjswanson I’d love contributions to the docs…Just fork fulcro’s github repo, and edit DevelopersGuide.adoc, then send a PR.

right on , will do

there is a nice plugin for IntelliJ that will let you preview as you edit, or you can use the Makefile to build the HTML (`make book`)

cool thanks, i'll check it out

output is docs/DevelopersGuilde.html

you have to have asciidoctor installed for make, which is ruby-based

IntelliJ plugin is just called Asciidoc, and works with the free version

:thumbsup:

anyone run into the issue with react where the cursor jumps to the end of an managed input? it prevents inserting text into the middle of a string, for example, because when the value changes it will re-render and the cursor position defaults to the end of the input field

looks like there may be workarounds to preserve and re-set cursor state before and after an update, but i'm wondering if there is a helper or a "fulcro way" to fix this

i still see one error for missing main.js , but everything looks fine anyways

main is needed if running from figwheel

so yes, that is ok

@eric.shao So, I figured out the bug. Load has an old option called :marker that historically defaults to “true”, but “true” means overwrite the target value with a load marker. I fixed it to give a warning on the case of special load targets (since the marker would overwrite the original vector), and turn the marker off. So, your original code should work with 2.5.9-SNAPSHOT.

Version 2.5.9-SNAPSHOT is on clojars with two load-related bug fixes. One has to do with load splitting/combining, and the other with load markers overwriting to-many relations when the load marker is boolean true (which is the default).