unparse fails with [org.clojure/clojurescript "1.7.122"] due to clojure.string/replace

[hipitihop]

Tested with com.andrewmcveigh/cljs-time "0.3.11" and clojurescript 1.7.58 and earlier and the following test passes, but fails in 1.7.122 due to clojure.string/replace changes.

(deftest default-basic-time
  (testing "Confirm base cljs-time :basic-time uses zero time, seconds & milliseconds"
    (let [formatter-no-time   (datef/formatters :basic-date)
          formatter-with-time (datef/formatters :basic-date-time)
          iso8601             "20150408"
          from-str            (datef/parse formatter-no-time iso8601)        ; instantiate UTC from string without time
          from-date           (datef/unparse formatter-with-time from-str)]  ; convert back to UTC string with time
      (is (instance? js/goog.date.UtcDateTime from-str))
      (is (= "20150408T000000.000Z" from-date) "We expect UTC time including milliseconds to be zero."))))

[malchmih]

+1

[levand]

+1

[andrewmcveigh]

Thanks for the report.

This should be "fixed" in ccb19d0 by using the old string/replace definition, until I merge in some bigger changes that shouldn't require string/replace anymore.

Released as 0.3.12.

[malchmih]

@andrewmcveigh Works for me, thank you!

[danielcompton]

I didn't fully understand the root cause of this issue. Is it a bug in cljs that should be reported?

[andrewmcveigh]

@danielcompton Not sure if it's bug or not, but it's a change in behaviour to align with jvm clojure[1].

I don't think this case is tested, so maybe it is a bug. I guess I'd need to better understand the motivation of the change to know if it's intentionally breaking this behaviour.

[1] clojure/clojurescript@3e14690

[adwelly]

I think I may have encountered something related to this:

Using [org.clojure/clojure "1.7.0"]
[org.clojure/clojurescript "1.7.122"]
[com.andrewmcveigh/cljs-time "0.3.14"]

using lein figwheel I can do the following in the repl

......(fighwheel launches)
Prompt will show when figwheel connects to your application
To quit, type: :cljs/quit
cljs.user=> (ns cljs.user (:require [cljs-time.format :as f][cljs-time.coerce :as c]))
;; This loads cljs-time format and cljs-time coerce, I can't seem to get (use ...) working.
cljs.user=> (def datefmt (f/formatter "dd-MM-yyyy")) ;; I create a custom formatter
#'cljs.user/datefmt
cljs.user=> (def atime (c/from-long 1400000000)) ;; and a datetime - sometime in 1970
#'cljs.user/atime
cljs.user=> atime
#object[Object 19700117T045320]
cljs.user=> (f/unparse datefmt atime)
#object[TypeError TypeError: Cannot read property 'call' of null] ;; <- something wrong
cljs.user=> (f/show-formatters)
#object[TypeError TypeError: Cannot read property 'call' of null] ;; <- show-formatters uses unparse too
cljs.user=>

I'm always open to the possibility that I'm doing something catastrophically daft, but this code was working with an earlier version of clojurescript a few months back.

[andrewmcveigh]

It's probably related to #41 & bhauman/lein-figwheel#210 rather than this issue.

What version of figwheel are you on?

[adwelly]

Just looked - figwheel 0.4.0, I see there's a figwheel 0.4.1 now. I shall give it a go and report back

Update - Yes, as you said, figwheel 0.4.0 is the root cause, the upgrade to 0.4.1 seems to have fixed the issue completely. Thanks very much for the prompt reply; it's very much appreciated.