Fork me on GitHub
#cljs-dev
<
2021-12-02
>
thheller05:12:09

@braden.shepherdson take a look at the contents of the closure-library.jar file. that is basically what it needs to find on the classpath

borkdude13:12:13

@alexmiller would you mind if I added the cognitect test runner to tools.namespace while doing TNS-51?

borkdude13:12:02

{:deps {org.clojure/tools.namespace {:local/root "." :deps/manifest :pom}}
 :aliases
 {:test
  {:extra-paths ["src/test/clojure"]
   :extra-deps {io.github.cognitect-labs/test-runner
                {:git/tag "v0.5.0" :git/sha "b3fd0d2"}}
   :exec-fn cognitect.test-runner.api/test
   :exec-args {:dirs ["src/test/clojure"]}}}}

Alex Miller (Clojure team)13:12:47

Nope, but we won't use that for any of the ci or release - that's all Maven

borkdude13:12:44

that's fine, I just prefer to run the tests like this :)

borkdude13:12:53

if I add a deps.edn like the above and someone uses tools.namespace via a gitlib, will it resolve via the pom via deps.edn?

borkdude13:12:26

I think that should be the desired behavior (not override the pom with the deps.edn)

Alex Miller (Clojure team)13:12:53

deps is preferred right now over pom

Noah Bogart14:12:10

looks like the indentation is different:

Noah Bogart14:12:15

maybe the result of tabs vs spaces?

borkdude14:12:53

This is what I'm seeing locally:

(keyword? form)  ; Some people write (:require ... :reload-all)
           nil
-       :else
+        (string? form) ; NPM dep, ignore
+          nil
+        :else

borkdude14:12:01

with git diff master

Noah Bogart14:12:16

hm! then maybe itโ€™s just jira being wacky. nevermind!

borkdude14:12:45

the whitespace is indented weirdly anyway, so it's very hard for my editor to not change anything about that

๐Ÿ‘ 1
borkdude13:12:26

@alexmiller what I mean is: I redirect to the pom in the deps.edn:

:deps {org.clojure/tools.namespace {:local/root "." :deps/manifest :pom}}

borkdude13:12:43

it works and the pom.xml is still the only source of truth regarding deps

Alex Miller (Clojure team)14:12:29

oh sorry, I didn't see all the parts of that question on my phone, seems cool

borkdude14:12:44

I'll submit the patch for this one: https://clojure.atlassian.net/browse/TNS-57 when TNS-51 has been processed, to not have to rebase patches.

borkdude14:12:11

Ffn JIRA, while I'm typing a name and selecting a person, it changes the UI from under me, so the wrong person gets assigned (and probably received a notification).

Alex Miller (Clojure team)14:12:12

feel free to file a jira with jira :)

Alex Miller (Clojure team)14:12:26

I probably won't look at the tns stuff until monday earliest

๐Ÿ‘ 1
borkdude20:12:59

I see you released a new tools.namespace.

Alex Miller (Clojure team)20:12:09

nothing from above, just deps bump

Alex Miller (Clojure team)20:12:42

not going to look at it today

borkdude20:12:00

ok, no hurry from my side

borkdude20:12:09

I'll just keep using the fork until you are ready

borkdude22:12:56

I noticed that ClojureScript depends on a fairly old version of clojure data json (0.2.6). Clojure.data.json has gotten significant performance improvements over the last few months, perhaps consider upgrading.

โ˜๏ธ 2
dominicm23:12:29

fwiw on the unshaded vs shaded compiler. The shaded compiler contains a copy of guava that's been prefixed with the google closure namespaces, so it doesn't conflict with the real copy of guava on your classpath. That could be an interim fix. I've found the problem in the build.bazel and let them know, but I'm not sure if there's any interest in actually fixing this from their end at this point.

๐Ÿ˜ž 1
Braden Shepherdson16:12:33

there are friends of Clojure(Script) at Google - how significant is the change needed?

dominicm11:12:41

It's not huge, just needs some bug fixes in their build process. https://groups.google.com/g/closure-compiler-discuss/c/5ofdcVdWgQs