Clojurians Log v2

I'm considering a .babashka/config.edn or babashka.edn where you can store things like :paths, :classpath-cmd (to calculate a classpath using e.g. clojure or using a bb script itself), :pods: pod specific settings, like where to download them. If you have any input on this, here is the issue: https://github.com/borkdude/babashka/issues/473

What would be the easiest way to craft something in bb that simply reads a number of edn forms from some file. I must be missing something obvious, but I invariably get off into weeds with needing to have a pushback-reader to use (clojure.edn/read). A bb (or clj) program that reads a file full of edn into memory should be hello world - so I must be missing somthing bigtime.

@hoppy so you have multiple values in one file, not like one value inside a vector right?

correct, and it's full of maps that are across multiple lines

@hoppy In one-liners you can do this:

$ echo '{:a 1} {:a 2}' | bb -I -e "*input*"
({:a 1} {:a 2})

yes, but I want to open the file from within the script

let me give you the variant that is suitable for scripts

$ echo '{:a 1} {:a 2}' | bb -e "(take-while #(not= ::eof %) (repeatedly #(edn/read {:eof ::eof} *in*)))"
({:a 1} {:a 2})

when reading from a file, replace *in* with (io/reader file)

I thought edn/read needs a pushback reader

oh right.

$ echo '{:a 1} {:a 2}' | bb -e "(let [reader  (java.io.PushbackReader. (io/reader *in*))] (take-while #(not= ::eof %) (repeatedly #(edn/read {:eof ::eof} reader))))"
({:a 1} {:a 2})

that's usually the sort of thing I come up with as well. Seems a little harsh, but ok.

I was considering making this a util function in babashka

it seems to be to be a missing bit in http://clojure.java.io frankly

<http://babashka.io|babashka.io>

I'm all for that as well

I had an issue for that here, but then decided to close it as documentation could maybe fix that better. https://github.com/borkdude/babashka/issues/613

but maybe http://babashka.io which will replace the I/O flags (that some people find confusing) may work

it's a tricky spot. putting things in bb that wouldn't be in jvm may cause hardship for peeps that are unknowing using bb as a gateway drug.

the way I solve that is publish every namespace as a JVM lib as well

e.g. babashka.curl and babashka.process are both JVM libs

even babashka.pods

ok, that's slick

we could also make one big babaskha.core or babashka.utils namespace where we accumulate functions that are missing from other core namespaces

not sure

dunno maybe I'm biased, but it seems like this should be a 1-liner in script space, at least if you would rather edn be the storage format than json or yaml. It's a little to bitey to get this loaded for my taste.

no matter how you do it, seems worth doing.

this is fun btw: https://grep.app/search?q=edn/read%20%7B%3Aeof%20

https://github.com/cldwalker/bb-clis/blob/453433e4a6d082d78751eb175b08bf3351481d9f/src/cldwalker/babashka/util.clj#L100-L105

we could make babashka.edn as a counterpart to clojure.edn and add missing functions. similar for http://clojure.java.io / http://babashka.io. even if those small nss don't contain a lot of functions, they could be bundles in a babashka.misc library that bundles these little namespaces?

<http://babashka.io|babashka.io>: push-back-reader
babashka.edn: edn-seq
;;=&gt;
(bedn/edn-seq (bio/push-back-reader (io/reader file)))

we could make this easier. and create the pushback reader for you if it isn't one.

(bedn/edn-seq (io/reader file))

(bedn/edn-seq file)

(defn edn-seq [reader-coercible]
  (let [reader (cond (instance? java.io.PushbackReader reader-coercible)
                     reader-coercible
                     (instance? java.io.Reader reader-coercible)
                     (java.io.PushbackReader. reader-coercible)
                     :else (java.io.PushbackReader. (io/reader reader-coercible)))
        sentinel (Object.)]
    (take-while #(not (identical? sentinel %))
                (repeatedly #(edn/read {:eof ::eof} reader)))))

https://grep.app/search?q=edn-seq

@hoppy At least I added docs here how to translate i/o flags to scripts: https://github.com/borkdude/babashka/blob/master/doc/io-flags.md

looks good!