Clojurians Log v2
cljs-dev
there is a Windows CI, thought it doesn't mean this would appear
minimal repo w/ just ClojureScript would be great
Will do
So repeatable but not new since it fails on 1.10.597 and 1.10.758. I can get it to consistently fail with :whitespace and :simple optimisations. Dev repl and compile, and :advanced optimisations is fine.
I'll raise an issue
@olivergeorge https://clojure.atlassian.net/browse/CLJS-2401
looks somewhat similar?
Yep seems the same. Hmm, perhaps I should close as a DUP.
Perhaps things have evolved a little since then :advanced seems to work. (Or my example was too simple - I’ll check this... still works with more complex example under :advanced)
Thanks for the link.
np, I vaguely remembered seeing java.nio.file.InvalidPathException: Illegal char <:> before somewhere 😛 but can't remember what I did in shadow-cljs to fix it
Yeah, I see your comments there. Do you still have the workaround in place?
https://github.com/google/closure-compiler/issues/2611#issuecomment-380722074
I think somewhere paths must be normalized so it doesn't write windows paths into JS output
yeah my hack is still around but was fixing a different issue with [synthetic:1] sources (which the closure compiler sometimes adds itself)
never checked if thats still needed though
Bug?
(into (transient []) [1 2 3]) ;; Error: No protocol method ICollection.-conj defined for type cljs.core/TransientVector: [object Object]
(implements? IEditableCollection (transient [])) ;; false
there's a similar ticket in clj I think
actually, I am misconnecting, although there are some old issue with the hierarchy for transient colls
I'm not sure how closely cljs matches clj in this area
@roman01la not a bug that doesn't work in Clojure either
in general please try it first in Clojure 🙂
@olivergeorge thanks
moving that one to blocker as well as some other old path bugs
feel free to chime in on other old ugly bugs that you believe should be fixed up for the next release
feature-wise the only thing I'll be looking at is the ns form stuff
@dnolen Interesting, but PersistentVector does implement IEditableCollection, how's that (implements? IEditableCollection (transient [])) results in false?
(transient []) is not a PersistentVector
IEditableCollection means "can be made into a transient"
you're asking if you can (transient (transient []))
ah right, thanks!
@alexmiller website could use a bump
https://groups.google.com/d/msg/closure-library-discuss/-znZL19uEnY/Ksugz4l0BQAJ
this is promising looks like Google is aware of the impact for Closure Compiler users
yeah I'm most curious about how they are going to do the namespacing system. I'd expect them to move to regular import/export for the TS sources. not sure if they are going to bother to keep the old namespaces in place since closure can consume import/export just fine
would it be possible in the future to have GCC consume foreign TS?
oh that’s literally what the message is about: don’t require GCC users to have to compile TS 🙃
@alexmiller hrm the ClojureScript website seems broken?
when I click on the links I'm getting file downloads now
anybody else seeing this?
which is a bit of a bummer; I found myself wanting to write some JS for some lib code, and decided to try out TS. I started thinking about how I would include it in a CLJS lib/app. Atm I guess the best option is to ship the compiled artifacts to NPM
I see this in Chrome & Safari
yup
same on chrome
and on ffox
“getting started” on the front page works, but the top nav links download a file
yeah, I'm working on it
don't even ask
don't experiment with stuff in prod
:)
should be fixed
I saw some references to TS in the GCC code. never actually tried it though.
maybe it is already capable of compiling TS code
thanks!
I guess there's a TS -> closure project called tsickle
Maybe I could compile the TS code to closure JS and then ship that in my jar with CLJS code