Clojurians Log v2

@borkdude as mentioned on twitter, how I compose steps in scripts…

this is the driving fn…

(defn run!
  [{:keys [verbose]} steps]
  (let [all-results (atom {})]
    (doseq [step steps]
      (let [{:keys [action success?]} step
            result (action @all-results)]
        (println)
        (println ">>" (:name step))
        (when verbose
          (pprint result))
        (if (success? result)
          (swap! all-results assoc (:name step) result)
          (throw (ex-info "action failed" result)))))
    (println)
    (println "Success!")))

and calling it….

(utils/run! options
            [{:name     :get-tag
              :action   utils/step-get-git-tag
              :success? (fn [result]
                          (not (nil? (:latest-commit-tag result))))}
             {:name     :get-sha
              :action   utils/step-get-git-sha
              :success? (constantly true)}
             {:name     :compress-main-js
              :action   (fn [_]
                          (utils/compress-file! "./public/client-js" "main.js"))
              :success? (fn [result]
                          (> (:size-compressed result) 0))}
             {:name     :deploy-editor-js
              :action   (fn [results]
                          (let [sha (get-in results [:get-sha :sha])]
                            (utils/deploy-compressed-asset-to-s3!
                              {:region       bucket-region
                               :content-type "application/javascript"}
                              "main.js.gz"
                              (str  "/editor/main." sha ".js"))))
              :success? (fn [result]
                          (nil? (:Error result)))}
             ])

Newbie question, if you have a deps.edn file next to a BB script how do you download the dependencies and make them available in the BB NRepl?

works great. the options come from tools.cli

one big seq of side-effects

Right now bb doesn't read your deps.edn, but you can pass it manually to (babashka.deps/add-deps (edn/read-string (slurp "deps.edn")))

Thanks. Added to https://github.com/babashka/babashka/issues/756

Anyone has babashka doing do some work with MS Excel files? For example using DocJure?

@marco.pasopas docjure depends on apache poi which is a java lib and you can't load that in babashka unfortunately.

writing a pod for it would be the way to get it in babashka reach (but this requires work)

Any other relevant lib that would enable me to convert excel into csv using babashka?? Maybe i am missing something. On the pod remark.. would a newbie be able to write a pod? Is it very complicated?

Would love to contribute in my spare time

It's fun but not that trivial maybe if you're just starting out with Clojure. Good learning experience though. Docs are here: https://github.com/babashka/pods Examples: https://github.com/babashka/pod-registry

The other option for going from excel to csv would be to install a binary who can do that for you and then shell out to it. csvkit has support for it: https://csvkit.readthedocs.io/en/latest/scripts/in2csv.html

@otfrom might also have some ideas how to do this in the shell

I do all my excel in poi (trying out fastexcel atm)

yeah, I've also used poi via docjure in the past. but this question was about how to do it from babashka. I think a poi (docjure or otherwise) pod might be nice for scripting.

but csvkit also works if you just want a csv from an excel file

So if I have a small script like this:

#!/usr/bin/env bb                                                                 
                                                                                  
(require '[clojure.string :as str])                                               
(require '[<http://clojure.java.io|clojure.java.io> :as io])                                               
                                                                                  
(defn read-lines [file]                                                           
  (with-open [r (io/reader file)]                                                 
    (doall (line-seq r))))                                                        
                                                                                  
(let [file *command-line-args*]                                                   
  (doseq [w (-&gt;&gt; (read-lines file)                                                
                 (str/lower-case)                                                 
                 (#(str/split % #"\W+"))                                          
                 (frequencies)                                                    
                 (sort-by val &gt;)                                                  
                 (map #(str (key %) " " (val %))))]                               
    (println w)))

I keep getting this error: ----- Error -------------------------------------------------------------------- Type: java.lang.IllegalArgumentException Message: Cannot open <("foo.txt")> as an InputStream.

@chase-lambert *command-line-args* is a seq, not a single string

ahh, ok, thank you. So putting (first *command-line-args*) works but what would you consider a more idiomatic way to feed in a text file?

that's how I do it a lot too

is there a shared library version of babashka?

@goomba you mean, native ?

ah yeah, right. many operating systems. Sorry lots of moving parts. Trying to figure out how to package babashka as a binary extension for Python

@goomba I have an example of that here: https://github.com/borkdude/sci/blob/master/doc/libsci.md

whaaaat

dude you rock

you will have to write your own API for this like I did for libsci of course

you did the hard part 😄 I can write a wrapper, hopefully

I have a variation of it here: https://github.com/borkdude/plsci

It's not a production quality thing but maybe useful as an example

The brainstorm continues:

$ bb :example 1 2 3
:args ("1" "2" "3")

$ cat bb.edn
{:paths ["."]
 :tasks {:example {:task/type :babashka
                   :args [:invoke utils/foo]}}}

$ cat utils.clj
(ns utils)

(defn foo [& args]
  (prn :args args))

:invoke would then be a built-in task like:

$ bb :invoke clojure.core/prn 123 456
"123" "456"

why not -X?

I find -X usually more verbose than just passing string arguments

Also '"foo"' for passing strings is something which I don't find esthetically pleasing really

It has no way to updating vectors in maps either. It's fairly limiting

updating vectors in maps? 😮

Say you have a preconfigured arg map as {:paths ["foo" "bar"]}. Now you want to add another path from the command line. No way to do that with -X

ah so you're saying something like bb :paths ["new-path"] :invoke blah

would do the trick?

no, you will just write bespoke command line parsing that makes sense for your function. And then you can invoke it with bb :your-task foo bar baz

or bb :invoke your-ns/your-fn foo bar baz, but the idea is to configure the :your-fn named task in bb.edn

similar to bb -m your-ns foo bar baz but now you can have arbitrary mains

we can add something like -X later, if that way of passing args becomes popular, but every time I see a question for which the answer is '"foo"' in #tools-deps I have some doubts about this :). Just being able to pass raw strings looks like an uncontroversial first step

@borkdude the current version of babashka/babashka.pods is not released to clojars, right?

there’s only a 0.0.1-snapshot release on clojars

could you make one release that’s not snapshot and is current?

my tests need it, thanks

or you expect people to use babashka/babashka instead?

ok, babashka/bashka seems to work.

I was just checking out a Cider issue for adding cider-jack-in support to Babashka: https://github.com/clojure-emacs/cider/issues/2947 . I have been suffixing my babashka scripts with a .bb extension to signify script's runtime (which would solve this issue and perhaps a related issue in clojure-lsp). Is this not common practice? For some reason, I thought it was, but it turns out that none of my coworkers add this suffix either!

@huahaiy Do you use leiningen for testing? With deps.edn you can use it as a git dep

But if needed I will make a non-snapshot release

yes i use lein test

It's not enforced by any means. I usually choose .clj to have good syntax highlighting by default

ok, can you post an issue at the pods repo to remind me?

sure

done https://github.com/babashka/pods/issues/31

ah, I've added

(add-to-list 'auto-mode-alist '("\\.bb\\'" . clojure-mode))

thx

@huahaiy Done, released

kondo doesn't warn on #!/usr/bin/env bb afaik

bb scripts can also be .cljc

and many times bb scripts are used on the JVM as well. So having .clj for both bb and clj is also better for that

so it's not that easy

imo cider-jack-in isn't worth it, I never use it myself. I always start a server in the terminal and then connect to it

having a development process as a child process of your editor is wrong imo

e.g. when you have a non-terminating loop in your dev process, you will have to kill your editor, or vice versa: if your editor crashes, your dev REPL is also lost

You are right about Kondo, sorry for the incorrect example. It is probably not worth it to make such a big design decision based solely off of jack-in, but I do think there might be value in distinguishing the runtime in the same way that clj, cljs, and cljr do (even if bb is just a subset of clj). If you are writing a script for multiple runtimes, then you can of course still use clj or cljc. Of course, you've thought everything through much more deeply then I have, so I am probably missing some other angle

People are free to use .bb if they want

This is also supported from the perspective of bb itself, when loading files from the classpath

yup! nothing more than a convention that I thought other people were using, but that I must have made up

some people do use it

and there is a built-in support for it

so go use it

Hi! I was looking to add support for the <http://java.io|java.io>.Serializable interface in my own build. I made https://github.com/frankitox/babashka/commit/de2f3a660e1d61d3616c851ef6c5a19d449c0185 modification without success, based on what I found about <http://java.io|java.io>.FileFilter , because that's an interface too. Am I missing something?

Note that <http://java.io|java.io>.Serializable is an old marker interface, from what I read in the https://docs.oracle.com/javase/7/docs/api/java/io/Serializable.html. I need it in order to use https://github.com/clj-commons/ring-buffer/blob/master/src/amalloy/ring_buffer.clj#L9, which could potentially allow me to use https://github.com/BrunoBonacci/mulog.

The way I tested if the compiled bb binary works is by doing something like:

cat test.clj
(ns test
  (:import [<http://java.io|java.io> Serializable]))
bb test.clj

So what didn't work?

Oh, just realized I forgot to run ./bb instead of bb , so was picking up the babashka binary in the path 🤦. Now it works 😀

BTW, thank you for the build and dev doc files, they were quite helpful!

trying to use babashka hiccup via

(require '[hiccup.core :as html])

@macrobartfast

$ bb -e "(require '[hiccup.core :as html]) (html/html [:a 1])"
"<a>1</a>"

did you mean :refer [html] ?

I think my mistake was that I was using

(html [:a 1])

(html/html [:a 1])

Checking that now.

yep! that was it. thank you!

(and :refer does the trick as well)