The following code calls copy source:

(defmethod copy-source*
  :jar
  [src dest options]
  (when-not (= :thin (:jar options))
    (when *verbose* (println src))
    (consume-jar (path src)
      (fn [inputstream ^JarEntry entry]
        (let [^String name (.getName entry)
              last-mod (.getLastModifiedTime entry)
              target (.resolve ^Path dest name)]
          (if (.isDirectory entry)
            (Files/createDirectories target (make-array FileAttribute 0))
            (do (Files/createDirectories (.getParent target) (make-array FileAttribute 0))
              (try
                (when (.startsWith name "META-INF/versions/")
                  (reset! multi-release? true))
                  ;; name is the filename in jar, inputstream is a JarInputStream
                (copy! name inputstream target last-mod) ;; note the inputstream here
                (catch Throwable t
                  (prn {:error "unable to copy file"
                        :name name
                        :exception (class t)
                        :message (.getMessage t)})
                  (swap! errors inc))))))))))

(defn consume-jar
  [^Path path f]
  (with-open [is (-> path
                     (Files/newInputStream (make-array OpenOption 0))
                     .BufferedInputStream.
                     JarInputStream.)]
    (loop []
      (when-let [entry (.getNextJarEntry is)]
        (f is entry)
        (recur)))))

(defn copy!
  ;; filename drives strategy
  [filename ^InputStream in ^Path target last-mod]
  (if-not (excluded? filename)
    (if (Files/exists target (make-array LinkOption 0))
      (clash filename in target)
      (do
        (Files/copy in target ^"[Ljava.nio.file.CopyOption;" copy-opts)
        (when last-mod
          (Files/setLastModifiedTime target last-mod))))
    (when *debug*
      (prn {:excluded filename}))))

Jarinpustreams are odd

A jarinputsream is, if I call, sort of a mix between and iterator and an inputstream

I was wondering if it copies according to what the getnextentry kinda does something weird to it

Yes, Everytime you call getnextentry, the jarinpustream sort of resets as the inputstream for the next entry

The zip file stuff (which is where I remember this from) works the same way

Just noticed this thread 🙂

cool. thanks @U0NCTKEV8

i kinda knew you would jump in except that you got beaten to the punch. thanks though @U04V70XH6

Yeah, I must admit, I was a bit puzzled by the original code but figured that it must behave along the lines that @U0NCTKEV8 explained, in that the input stream is "chunked" to match the files in the JAR file, so a single input stream can be used across multiple logical files...

I'm curious @U11BV7MTK what made you dig into the source?

yeah i was surprised that the entries aren't JarEntryInputStreams or something similar.

i've never been heavy interop or know jar related bits so what better way to refresh than by reading ghadi code

@dominicm, you can use membrane, https://github.com/phronmophobic/membrane

see https://phronmophobic.github.io/membrane/api/membrane.skia.html#var-draw-to-image.21

here’s an example:

(ns membrane.example.todo
  (:require [membrane.skia :as skia]
            [membrane.ui :as ui
             :refer [image
                     translate
                     label]])
  (:gen-class))

(defn overlay []
  (let [img (ui/image "resources/Clojure_logo.svg.png"
                      [400 nil])]
    [img
     (translate 50 50
                (ui/label "This is a clojure logo"
                          (ui/font "/Library/Fonts/Comic Sans MS.ttf"
                                   45)))]))

(defn -main [& args]
  (skia/draw-to-image! "myoverlay.png" (overlay)))

https://github.com/phronmophobic/membrane/blob/master/project.clj#L7-L30 that's a lot of dependencies - a surprising set really. I don't think figwheel should be in there?

you’re* right

it should not

Will take a look though, it'd be nice to have cljs support too.

It's not necessary, we have no specific plans, but it's a nice idea.

i’m not sure what you mean by cljs support

but there is a graphics backend for webgl

That's what I mean :). The java2d recommendation doesn't have cljs support (obviously). So using a library that does is positive (like yours).

😄

feel free to ping me if you have any questions. feedback is welcome!

fyi, just pushed a new version to clojars [com.phronemophobic/membrane "0.9.9-beta-SNAPSHOT"] that moves figwheel to dev dependencies.

Cool, thanks :) I think we'll get to this next week.

that looks handy, definitely an option I hadn't seen.

The exit status of the clojure command will be 0 for success and non-zero for failing tests. You can check $? in the shell.

Some CI/CD systems handle that automatically for you (and abort when any command exits with a non-zero status).

GitHub Actions running the CLI tools: https://github.com/seancorfield/next-jdbc/blob/master/.github/workflows/test.yml

CircleCI running the CLI tools: https://github.com/seancorfield/next-jdbc/blob/master/.circleci/config.yml

(in both cases, the abort is automatic on test failures)

Thank you @U04V70XH6! I'll take a look at this =]

@U04V70XH6 for me $? is always 0 after test failures in my terminal. We are using Jenkins and Groovy to run our tests. Right now I was using groovy's sh command to turn the test command and that is returning 0 for me on failures. Any ideas for me in this environment?

I'm not the most familiar with Jenkins so I may be getting something wrong. But CircleCI and Github Actions seem like competitors to Jenkins rather than a plugin I could be using here

@UN1E6R822 Are you using an up-to-date version of the Cognitect test-runner? Early versions did not set the exit status on failure.

(! 662)-> clj -A:new lib brian/test-failure
Generating a project called test-failure based on the 'lib' template.
The lib template is intended for library projects, not applications.

Fri May 15 15:12:19
(sean)-(jobs:0)-(~/clojure)
(! 663)-> cd test-failure/

Fri May 15 15:12:23
(sean)-(jobs:0)-(~/clojure/test-failure)
(! 664)-> clojure -A:test:runner

Running tests in #{"test"}

Testing brian.test-failure-test

FAIL in (a-test) (test_failure_test.clj:7)
FIXME, I fail.
expected: (= 0 1)
  actual: (not (= 0 1))

Ran 1 tests containing 1 assertions.
1 failures, 0 errors.

Fri May 15 15:12:34
(sean)-(jobs:0)-(~/clojure/test-failure)
(! 665)-> echo $?
1

(! 666)-> cat deps.edn 
{:paths ["src" "resources"]
 :deps {org.clojure/clojure {:mvn/version "1.10.1"}}
 :aliases
 {:test {:extra-paths ["test"]
         :extra-deps {org.clojure/test.check {:mvn/version "1.0.0"}}}
  :runner
  {:extra-deps {com.cognitect/test-runner
                {:git/url ""
                 :sha "f7ef16dc3b8332b0d77bc0274578ad5270fbfedd"}}
   :main-opts ["-m" "cognitect.test-runner"
               "-d" "test"]}
  :jar {:extra-deps {seancorfield/depstar {:mvn/version "1.0.94"}}
        :main-opts ["-m" "hf.depstar.jar" "test-failure.jar"]}
  :install {:extra-deps {deps-deploy {:mvn/version "0.0.9"}}
            :main-opts ["-m" "deps-deploy.deps-deploy" "install" "test-failure.jar"]}
  :deploy {:extra-deps {deps-deploy {:mvn/version "0.0.9"}}
           :main-opts ["-m" "deps-deploy.deps-deploy" "deploy" "test-failure.jar"]}}}

Likely the groovy process isn't exiting with the same status as the clojure process it shells out to

I would just use, uh, I think Jenkins calls them freestyle projects? It has been a few years

Having groovy in the mix is just another thing to go wrong

The docs seem to claim on the latest version of the groovy stuff and non-0 statues will cause it to throw an exception

(some-> % float) should work.

thanks 🙂

also there are few reasons to use float rather than double

I guess I can use double thanks 🙂