This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
2017-07-30
Channels
- # aleph (4)
- # beginners (24)
- # boot (15)
- # cider (4)
- # cljs-dev (37)
- # clojure (73)
- # clojure-losangeles (1)
- # clojure-serbia (1)
- # clojure-spec (27)
- # clojurescript (78)
- # core-logic (3)
- # datascript (9)
- # datomic (10)
- # events (1)
- # lein-figwheel (1)
- # lumo (2)
- # off-topic (14)
- # om (6)
- # om-next (1)
- # parinfer (18)
- # pedestal (2)
- # protorepl (4)
- # re-frame (2)
- # reagent (56)
- # specter (6)
- # unrepl (2)
OK fixed CLJS-2281: https://dev.clojure.org/jira/browse/CLJS-2281
the attached repro had weird bugs (in the repro itself) but importing and using material’s textfield works
I think 854 has reintroduced a "relative path" issue on Windows again. I'll look into this on Monday.
@olivergeorge would be good to have it reported and fixed before the actual release then. I added some tests for relative-path
so we probably need more
maybe this is the culprit? https://github.com/clojure/clojurescript/commit/b88d290f95f6ac9bc7040ceaf8cfb3a57e6b9935
I did some cursory testing on my windows VM and didn’t see any issues
I'll do my best to check it out tonight then. +5 hours or so.
Confirming that the example from CLJS-2036 was fixed in 1.9.671 but fails again in 1.9.854
I'll prepare a bug and investigate but if anyone is interested here's a quick repo (lein mies plus cljs-2036 steps to reproduce).
Update project.clj to set the clojurescript version and run scripts\build.bat
on windows.
Includes a patch which fixes things (for me) and seems sensible.
@anmonteiro I think you were right about the code. Not sure about the tests... seems they didn't catch the platform specific nature of File/separator
@olivergeorge patch needs test
I’m thinking that if we decide in favor of @thheller ‘s suggestion of always enabling AMD processing, we can greatly simplify process-js-modules
like, we don’t need to know the :module-type
anymore, or even have the multimethod
if the Closure Compiler is smart enough to figure all those things out, then I think it also eliminates the problem described in this comment: https://github.com/clojure/clojurescript/blob/master/src/main/clojure/cljs/closure.clj#L2340-L2343
I’m going to play around with it
Hi Guys, I have a small window to help with CLJS-2285 but I'm not sure about testing and windows.
Is this the right refererence: https://clojurescript.org/community/running-tests
I think perhaps it's not updated for node/npm
95 tests, 32 failures running from Windows
Actually, I can work with it. The last error is a relative-name report which I think shows the bug I was reporting on master.
@dnolen With that patch there's one less error reported by lein test
.
perhaps we need File/separator bound to a dynamic variable to allow testing of windows behaviour without Windows running.
Yeah, this file path manipulation stuff is the devil. @anmonteiro tests highlight another issue.
I saw some weird things in that function which is why I added the tests
Thanks for that. It's nice that we can see them at test time.
These functions are impure and surprising due to the platform specific calls to java functions. Should we create some interop helpers which smooths out the inconsistencies? It'd be nice to deal in data (e.g. path as vector of strings).
sure. we had too many regressions because of that
^ this was answering the previous statement
not sure about helpers, better ask David
Gotcha
It might be a whole new can of worms. Only a gut feeling.
Just found to-path and path-seq. Will see if I that suits.