2.0.0-SNAPSHOT updated on clojars. History refined. There is a feature/2.0 branch of todomvc that works with the new version and can show a couple of the features of the new history support (compressed transacts, tx tracking)

@wilkerlucio So, the factory “fix” (apply) does break things a bit. It changes how children appear

When making defui components that expect children, they used to always appear as a vector, whereas not if there is just one, for example, then children will no longer be a vector…just an element. This will break existing code for 2.0. Not sure which way to go…breakage of existing code, or elimination of an annoying react warning

I think it might be fixable by changing how the prim/children function works to this:

(defn children
  "Returns the component's children."
  [component]
  (let [cs #?(:clj (:children component)
              :cljs (.. component -props -children))]
    (if (coll? cs) cs [cs])))

that way existing users that are pulling children always get a collection (as before), but the factory change can stay in place for when React pulls them

Using Fulcro 2.0. The :started-callback function has app as a parameter. But an assert on (prim/reconciler? (:reconciler app)) is failing.

weird

let me look…

you sure?

dump app itself…is it an atom or something?

function (config,state,meta,extmap,hash){ this.config = config; this.state = state; this.meta = meta; this.extmap = extmap; this.hash = hash; this.cljs$lang$protocol_mask$partition0$ = 2229700362; this.cljs$lang$protocol_mask$partition1$ = 139264; }

That was from (type reconciler)

not using devtools?

Well I thought I was, on Chromium? not Chrome.

I’m not sure why it would be working, but would lose its type

the code looks right to me, and most of that hasn’t changed…just my call to the construction of the Reconciler now uses mine, but it implements the protocol

and a lot of internal asserts on reconciler? are working

I'll clean everything again...

make darn sure you don’t have Om Next around

if you accidentally end up with both, things might compile and give you strange errors because some of your namespaces might have gotten missed or something

I will invalidate caches on IntelliJ this time.

I guess none of the internal asserts are pulling reconciler from the app. It is in the app for your external purposes…so it could be broken I guess. I just don’t see how

Yeah, I’m running a 2.0 devguide, and started callback gets a proper reconciler

type returns Reconciler

My bad, still had [org.omcljs/om "1.0.0-beta1"] in the project file.

yeah, I’m guessing you’ve got some compiler errors now

probably were still pulling in some Om code somewhere

My renaming stuff is naive…it assumes you followed the pattern of requires [om.next :as om]

if you did it any other way, it won’t find them

My misconception before was that namespaces will change with Fulcro 2, so it is more than a fork, which might keep the same namespaces??

No compile errors. I think fulcrologic/fulcro-spec and fulcrologic/fulcro-datomic also ought to be upgraded to 2.0 - they might be pulling in om??

spec has a 2.0 snapshot

fulcro-datomic, unfortunately, does not yet

at least I don’t remember making one….maybe I did

hold on…I’m using that in todomvc, and I got that working today, so it must be ok

fulcro-datomic has no dep on Om or Fulcro

it is standalone

so bump spec to 2.0.0-SNAPSHOT and you should be ok

make sure to scan lein deps :tree for Om, though

Yes, keeping "1.0.0-SNAPSHOT" for fulcro-datomic

getting compile errors now thanks 🙂

you saw the renaming guide in the new readme, right?

I supply a script

It is not a fork, per se, no. It is a completely new thing that keeps the same function names, but does create new namespaces for them. This is because they are technically different things.

so, the code looks pretty much the same, except for the requires

https://github.com/fulcrologic/fulcro/blob/2.0/README-fulcro-2.0.adoc

I'll look at the renaming guide now.

yeah, if you copy/paste that (assuming OSX or Linux) it should get you most of the way there

might also work in something like cygwin for winders….is that even a thing, or did they make something new…so long since I’ve touched windows…like since version 7 or something

I'm happily on Linux, left it at Windows 7 too. Never seen Windows 10!

curious if you use perl for that kind of stream editing, or something else. sed is ok but I really prefer perl regex support 😉

I used to use sed for things. Been a very long time since needed to do shell scripting. My professional existence has been very much Windows for last 7 years.

`Call to #'fulcro.cli … ailure :instrument↵" Call to #'fulcro.client.impl.data-fetch/place-load-markers did not conform to spec:↵ In: [1 0 :fulcro.client.primitives/ident] val: nil fails spec: :fulcro.client.primitives/ident at: [:args :items :fulcro.client.primitives/ident] predicate: ident?↵`

I changed the spec for ident to s/or nil? to get by that one.

Next is a stack trace of many Uncaught TypeError: Cannot read property 'call' of undefined, all for factory calls, starting with my TopRouter then going down to DropdownItem (bootstrap3 line 517) and finally ShadowDOM (elements line 82).

Problem happens when doing the swap! that mounts the Root component: No protocol method ISwap.-swap! defined for type undefined.

Hey @cjmurphy So, instrument should not be turned on just yet I think, if you’re turning it on (for spec)

that is strange…are you building thedev guide?

I’ve started adding more specs, but I’m not enforcing them yet because there is a bit of refactoring to do to get them to all pass via instrument

@tony.kay: I never tried building the dev guide, just went straight to my application. The culprit for why specs would be turned on is in my user.cljs - I added a (ts/instrument) in there then promptly forgot about it 😜