This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
2017-06-13
Channels
- # beginners (78)
- # boot (27)
- # cider (13)
- # cljs-dev (41)
- # cljsrn (4)
- # clojure (216)
- # clojure-android (1)
- # clojure-conj (6)
- # clojure-greece (1)
- # clojure-italy (11)
- # clojure-russia (127)
- # clojure-spec (63)
- # clojure-uk (34)
- # clojurescript (68)
- # core-async (5)
- # cursive (5)
- # data-science (1)
- # datomic (4)
- # dirac (11)
- # editors (7)
- # events (1)
- # graphql (12)
- # hoplon (39)
- # jobs (1)
- # liberator (3)
- # lumo (101)
- # off-topic (14)
- # om (3)
- # onyx (3)
- # parinfer (14)
- # re-frame (10)
- # reagent (2)
- # remote-jobs (1)
- # ring-swagger (17)
- # sql (21)
- # untangled (38)
- # vim (3)
- # yada (23)
Not sure if anyone has looked at Zach Tellman's https://github.com/ztellman/collection-check? Seems like that might be a good tester for the new hashing?
Can someone comment on this please: https://dev.clojure.org/jira/browse/CLJS-2036?focusedCommentId=45826&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-45826
Feels like I'm close to being able to fix but I'm not familiar with the correct approach for normalising paths (e.g. windows \\ and linux /)
Specifically (System/getProperty "user.dir")
returns "C:\\Repos\\utas-webapps\\frontend\\"
but I need "/C:/Repos/utas-webapps/frontend/"
in order to strip the prefix.
Or perhaps it's the url arg which is technically the bug in this case.
@favila I mentioned that all callers of equiv are IKVReduce in patch, but was probably did make that very clear. Yes the fallback is bad - that is indeed an undefined variable i introduced. Ouch! @theller the records no longer use equiv-map it uses a set of fields checks and then a regular = check on the extmap as of cljs-2079. Also, the call made to 'equiv-map' by PAMs are redundant, it is made in the else branch of the if statement: (if (map? other) (not (record? other)) .... (equiv-map coll other)) which is pointless as 'other' is known not be a map by this point so equiv-map will always return false.
I see my typos was fixed. Thanks @dnolen ! btw I can make a test case for the fallback case if it would be helpful.
I remember calling private functions used to be an error. Eg, I once had to hack around it like (js/cljs.core.fn__GT_comparator ...)
. Did that change?
@olivergeorge thanks for the clarifications on the ticket, very helpful
dnolen: Thanks! I'll do that.
I'm still not clear on why the bug is exhibited specifically when using :preload which leads me to think there's something more to it. I'm struggling to find that and don't have time just now to get more familiar.
This fix should address the immediate issue and seems safe.
@tmulvaney sure, go for it
Error warning: https://dev.clojure.org/jira/browse/CLJS-1702
Hmm, I wonder where I got the error from. I could've sworn it didn't work using a normal (fn->comparator ...)
call. Oh well...
@olivergeorge your patch looks OK but you need to follow the patch guidelines here https://clojurescript.org/community/patches
PSA: If you are authoring patches, you can now also include generative tests (using test.check
). The first such test has landed, exercising the ability for this to properly span JVM and self-hosted.
should there be a separate test workflow for those? generative tests take much longer to run
Instances defspec
seems to argue that Clojure hasn’t yet started to worry about test run time. (They are mixed in and there aren’t very many yet.)
We are seeing 15-30% decrease in total script time after CLJS-2079 and CLJS-1977 from Chrome's performance tab.
I know we have a strange use case, but this is a big deal for us. Kudos all around.
@dspiteself sweet!
the Clojure JVM build has both unit tests and generative tests (actually these are the old test.generative, not test.check) and it’s possible to run them independently or together. I usually run them both as the generative ones don’t actually take that long right now.
we also have test.check tests but they are currently mingled into the unit test segment (because reasons)
@alexmiller thanks good to know
Hi All. Possibly silly question. What's the recommended approach to building the clojurescript compiler on Windows? I used a docker vm to avoid the issue but perhaps there's a less fiddly way.
I get errors running script/build using the Git bash install on windows.
Exception in thread "main" java.io.FileNotFoundException: Could not locate cljs/closure__init.class or cljs/closure.clj on classpath., compiling:(C:\repos\clojurescript\script\aot.clj:1:1)
In case it's of interest this was my docker work around
docker run -v c:\repos\clojurescript\:/app -w /app -it clojure
maybe you could try to run it in linux subsystem in windows (I have never tried that before)
Interesting idea.
I wonder if that's a likely future standard approach or more a novelty.