add dataify/nav impl to Object that looks for the magic react element symbol?

when using clj 1.10.x, what sort of expectation should you have concerning what version of cljs to use? should you only expect/assume it to work with 1.10.x cljs?

I believe a pattern I’ve seen is that cljs 1.10 stuff seems to begin to embrace clj 1.10 additions, but cljs 1.9 wouldn’t do that

but it is also the case that I can use the latest clj 1.10 (RC) release with lower cljs versions at least back into the 1.9's. So I just am not sure what sort of upgrade flow is expected to be followed

Side note: not sure if I should have thi son #cljs-dev or #clojurescript - it may be the latter - if so, sorry

@mikerod ClojureScript currently depends on Clojure 1.8.0

I’m confused how to match up compatible numbers across the 2

in both directions

eg. I jumped a project to Clojure 1.10.0-RC5 to try it out, had an old ClojureScript v1.7.x - and it failed with spec issues (not super surprised)

but then I tried a ClojureScript v1.9.x and it was fine

but in general, I don’t know what is expected to be a combination that should work

I guess you’re saying ClojureScript “currently” requires Clojure 1.8.0+

That partially addresses it in one direction for me. Just was trying to have a better feel for the relationship. Perhaps there isn’t a really direct one. I tend to just dig around in cljs changes.md to see if something seems to depend on something of a more recent clj or have been fixed to be compatible with some newer clj

@mikerod I think expecting alpha features to work across releases is unlikely to ever lead to good results

other than that there is no expectation beyond aggregation only

Interesting

I think macro-spec breaking stuff was sort of a rare case

made newer clj not be able to compile older cljs

we don't believe in semantic versioning

all 1.10.X tells you is that

so there is some barrier there now

won't break backwards for a lot of the time

and you'll get Clojure 1.10 features (some of which may be under development, so proceed with some caution)

> we don’t believe in semantic versioning Yeah, I’ve heard the Rich talk on it. I’m in agreement with the points presented there.

So I guess the thing that is weird is the breaking newer clj that happened due to specs

but when that happens, you just have to find the incompatibility point

it's actually more subtle than that

Clojure may close thing that were underspecified which people knowingly or unknowingly abused

in other cases, ClojureScript may have a few additional patterns that don't exist in Clojure etc.

so there's a some grey areas - but c'est la vie

Fair enough. Thanks for providing the clarifications.

just a heads up on http://clojurescript.org … the cloud CI product we use for site builds is being turned off this weekend and so the auto-builds will not exist anymore. I am working on a replacement, but it will probably not be ready immediately. I am able to build and deploy either site locally so I can do it on demand, but you will need to poke me to do so. Sorry for the inconvenience! I’ve known about this for a while but of course it’s never urgent until it’s actually urgent… :)

Thanks for the heads up!

digging more into datafy. I think this line is supposed to check if we can add meta data to the value from datafy, but it seems to not work as I expect:

https://github.com/clojure/clojurescript/blob/master/src/main/cljs/clojure/datafy.cljs#L24

ex from Clojure:

user=> (def x (with-meta [1 2 3] {`clojure.core.protocols/datafy (fn [_] [2 3 4])}))
#'user/x
user=> (d/datafy x)
[2 3 4]
user=> (meta (d/datafy x))
#:clojure.datafy{:obj [1 2 3], :class clojure.lang.PersistentVector}

(def x (with-meta [1 2 3] {:clojure.datafy/datafy (fn [_] [2 3 4])}))

(d/datafy x)
;; => nil

(meta (d/datafy x))
;; => nil

there’s also the case where the datafied value is a JS object and it throws:

(def x (with-meta [1 2 3] {:clojure.datafy/datafy (fn [_] #js {})}))

(d/datafy x)
;; => Error: No protocol method IWithMeta.-with-meta defined for type object: [object Object]

I’ll create an issue for this. I think if I create patch for it I could probably also try my hand at addressing CLJS-2999

@lilactown first question, is what does the Clojure implementation do

before opening issues around this probably useful to have a few back and forths here first

ack, apologies

the Clojure implementation seems to work fine with arbitrary Java objects AFAICT:

user=> (def x (with-meta [1 2 3] {`clojure.core.protocols/datafy (fn [_] (Object.))}))
#'user/x
user=> (d/datafy x)
#object[java.lang.Object 0x3381b4fc "java.lang.Object@3381b4fc"]
user=> (meta (d/datafy x))
nil

the current Clojure impl uses this to check whether to add meta to the datafied value: (instance? clojure.lang.IObj v)

@lilactown link to the relevant source?

https://github.com/clojure/clojure/blob/master/src/clj/clojure/datafy.clj#L26

thanks

Yeah, I probably mis-ported that check as a call to object?

@lilactown proceed!

My hunch is that it probably should have been (satisfies? IWithMeta v)

@mfikes I think in this case it should be implements?

I think the instance? in the Clojure implementation is intentional

Agreed

@lilactown ^

thanks 😄 I’ll probably spend the day figuring out the contribution process

@lilactown it's really not so bad, did you submit your CA already?

just did

sweet!

@lilactown https://clojurescript.org/community/patches

the rest is pretty boring

make a defect ticket, etc.

awesome! thx 🙂 I created CLJS-3010

thanks!

@dnolen can confirm CA receipt

thanks!

how do I run tests in src/test/cljs/clojure/?

@lilactown there are scripts in script if I remember correctly, or you bootstrap a REPL

@lilactown See https://clojurescript.org/community/running-tests

folks, can I do something else in order to have the breaking change release note addition merged in? https://github.com/clojure/clojurescript-site/pull/279