Clojurians Log v2

Here is a minimal sort of bb.edn to give you an example:

{:paths ["scripts"]
 :deps  {io.github.paintparty/bling {:mvn/version "0.4.2"}}
 :tasks {
         build-assets {:task (exec 'assets/.-main)}
         datomic      {:task (exec 'datomic/-main)}}}

Sure, you can put your code in a library and just use the library. You still have to hook up the library to the task names though

So in addition to bling you would have another library

{:deps  {io.github.paintparty/bling {:mvn/version "0.4.2"}
         myns/shared-scripts        {:mvn/version "0.0.0"}}
 :tasks {
         build-assets {:task ??}
         datomic      {:task ??}}}

yeah, you don't have to put it on mvn though. it can be a local/root or git/url+git/sha lib

can the :task value can just be normal clojure code like :build-assets {:task shared-scripts.assets/-main} ?

yes

ah nifty! I assumed it was something special I guess. Is there a way to provide the :task and :doc from the library?

no, unfortunately not, although doc will be derived from the var I think, if you just use the symbol

:build-assets shared-scripts.assets/task-def ? which is something like (def task-def {:task ... :doc ...})

the task names are always static

gotcha, ok, that is probably good enough, the key is that something useful shows up in bb tasks

thanks @borkdude I will give this a spin

This works quite well!

One thing I might like to do is use data defed in the bb.edn file:

{:deps  {mylib/tasks {:local/root "../../some/path/tasks"}}
 :tasks {:init       (do
                       (def css-build-opts {:input  "resources/public/main.css"
                                            :output "resources/public/compiled.css"}))
         asset-build mylib.tasks/css-main ;; <-- this function gets cli args but cannot access css-build-opts

         asset-build2 {:task (mylib.tasks/css-main css-bild-opts *command-line-args*)
                       :doc  "Have to manually copy the docstring"}}}

now you only have to copy your bb.edn instead of also your code

fair enough

Do you have any plans to support dynamic tasks or loading tasks from deps?

I consider it a feature that all tasks are statically known for people glancing the bb.edn (similar to aliases in deps.edn )

I have:

#!/usr/bin/env bb

(ns t
  (:require [babashka.process :as p]
            [babashka.fs :as fs]))

(def dir (fs/parent (fs/parent *file*)))

(defn compress
  []
  (let [data-dir (

user=> (apropos "path")
(babashka.classpath/add-classpath babashka.classpath/get-classpath babashka.classpath/split-classpath babashka.fs/exec-paths babashka.fs/path babashka.fs/path-separator babashka.fs/real-path babashka.fs/split-paths babashka.wait/wait-for-path clojure.core/*compile-path* clojure.core/*source-path*  clojure.zip/path selmer.parser/known-variable-paths selmer.parser/set-resource-path!)
user=> (doc babashka.fs/path)
-------------------------
babashka.fs/path
([f] [parent child] [parent child & more])
  Coerces f into a Path. Multiple-arg versions treat the first argument as
  parent and subsequent args as children relative to the parent.
nil
user=> (babashka.fs/path "/tmp" "trace.snowflake")
#object[sun.nio.fs.UnixPath 0x4335fbfd "/tmp/trace.snowflake"]
user=>

ask the runtime for things related to paths, get a docstring of one that looks promising, check it in the repl

(let [data-dir (fs/file dir "_data")]
  ...)

Oh. Thanks!

(i truly think apropos and doc are phenomenal tools that are default in the runtime)

I was looking at https://github.com/babashka/fs/blob/master/API.md#babashka.fs/path and the description at the top ("Coerces f into a Path.") did not hint at that at all.

oh you mean multiple args to fs/path?

Yes, I completely missed that.

does that too. Probably I didn't mention it because I was so used to it. Let's fix that right away. Thanks for trying bb :)

I never knew that about either 😮

Do you want a PR or is it easier for you to fix directly?

yeah. much better than string wrangling with "/" , more portable between OSes too

Oh PR would be sweet!

Are the docs generated somehow or is editing the md file directly the right approach?

https://github.com/babashka/fs/pull/149

docs are generated from the var docstrings using bb quickdoc

👀

so the docs are in the docstrings in the code

and then extracted and frobnicated into a .md file

tl;dr: just update the docstring in the code, that will do, don't worry about anything else

Yep, following that. So the information I was missing is in the docstring, but not in the first sentence, which is why it wasn't at the top.

oh...

Because quickdoc https://github.com/borkdude/quickdoc/blob/main/src/quickdoc/impl.clj#L37 for the summary at the top.

a little too quick

I've updated my PR.

Much better than it was. I think we can now delete the second sentence, do you agree?

I think I would be fine without it, but it does add more precision, so really up to you.

Happy to remove it in my PR if you think that's the right call.

I'll merge it, you've been through enough as a fresh bb scripter :)

✅

Thanks!

Yes search the babashka book for Python, afk now :)

Got it, thanks! I'd searched for "shebang", didn't think of searching for Python 🙂

http://book.babashka.org it’s somewhere in there

Yes, I have found it:

(= *file* (System/getProperty "babashka.file"))