Clojurians Log v2

is bb doc not supported? tasks are looking good. depends are looking good, and bb doc should show depends!

Not yet for tasks, I will fix that soon

I've played with 0.3.1 before the weekend and came up with this kind of pattern:

bb -m my.runner/run some.ns/fn :a b :c 123 :d '#{:e :f "g" [:h]}'

Since I'm wrapping the babashka main functionality, I could add some conveniences: if you pass an unqualified symbol as the first argument, it assumes you're calling a function in the tasks ns; and I also have a my.runner/doc that is similar to bb doc and a my.runner/echo which is kind of a --dry-run debug

I'm wondering if the parsing of CLI arguments as a map of keyword/edn-values may be popular enough to be supported more directly by a babashka helper (ala process/tokenize ). Either way, just giving a positive experience report -- and bb.edn is very helpful! 🙂

@pithyless Interesting! the clojure CLI also supports passing map args as EDN, was that your inspiration?

yes

actually, I was hoping there was some code I could call directly, but my brief search was not fruitful

so I ended up heavily borrowing the ideas from this post: https://vlaaad.github.io/tools-cli-in-10-lines-of-code

I actually considered the idea of "default args", "command args", and "env args" could all be considered the same thing, if we try to map the CLI to the Clojure idea of flat maps with fully-qualified keywords. So maybe instead of: SOME_ENV=... command --verbose --some=args we could just say runner some.ns/command :some.env/stuff :value1 :some.cmd/arg :value2 :verbose true

and I could imagine that we could just merge mutiple maps (maybe part of it is just pulled from some file/env/wherever)?

Hi there I have two babashka scripts that parse lines of edn from stdin, is there any practical difference between using a straight -f vs using a main with -m?

(it's a migration script)

it seems from the manual that running a uberjar would not allow me to call two different namespaces/files?

@richiardiandrea I don't recommend using the -i -O flags with scripts really, they were mostly designed for one-liners. Just use edn/read if you need to read multiple lines of edn for example.

I documented here: https://book.babashka.org/#_scripts what the scripting alternatives are for the i/o flags

There is not really a practical difference between -f and -m. The main difference is that with -m the function you are invoking is explicit and with -f it's implicit via side effects

Just wanted to process things lazily from the shell piping as they come

When using an uberjar the main function must be set when creating the uberjar

or you can invoke it like this: bb -cp foo.jar -m foo

You can still do that, using *in*

@richiardiandrea

$ echo '1 2 3' | bb -e '(take-while #(not= % ::EOF) (repeatedly #(edn/read {:eof ::EOF} *in*)))'
(1 2 3)

that is a lazy seq of edn vals from stdin

Yes that's what I am doing and I have a def at the top level for that. So do one-liners need them?

@borkdude sorry I also had this side question 😃 I have.inspected the code and depstar only triggers one main. I basically need to call two mains 😃 in two different namespaces

I don't understand your question. If you use the above construction, you don't need any special flags

I think you would need to create one main function that combines the two

yeah that is not possible cause the two are piped one to the other in the script...can I do something like bb my.jar -m my-foo.main and then bb my.jar -m my-bar.main at the moment?

you can do bb -cp my.jar -m my-foo.main

so bb -cp my.jar -m foo/main1 | bb -cp my.jar -m foo/main1

but why use a pipe between two functions if you can just process the data within one process?

because I want to expose the intermediate result and possibly re-run it

it's some form of aid for my operators

the intermediate result can be inspected for problems, solve them, then run against it multiple times

thank you I will try your solution, it looks like it might work

never mind, I wanted to just understand when I need -IO if at any point I can use *in*

what do these flags do under the hood basically, but I am checking the source right now no worries

the flags i and I influence the value of *input* and -o or -O influence how results get printed

oh I see, so -o/`-O` is basically only useful if I don't println/`prn` myself and the script ends evaluating to a data structure...gotcha that makes sense

correct

well no it does not seem to work... This is what I see with jar -tf my-jar

bb.edn
src/
src/image_migration/
src/image_migration/insert_images.bb
src/image_migration/csv_to_image_todos.bb

----- Error --------------------------------------------------------------------
Type:     java.lang.Exception
Message:  Could not find namespace: image-migration.csv-to-image-todos.
Location: <expr>:1:10

does $uberjar_path have the correct value?

@richiardiandrea I see, the expected directory structure here is image_migration/insert-images.bb, not with src prefixed

so I think something went wrong in your uberjar production

uhm ok let me see there

bb.edn is

{:paths ["src" "resources"]}

ah I see... there is a bug here. the uberjar task does not yet take into account bb.edn :/

I logged an issue for this

ok glad we found it

so for now you will have to do it like this:

bb -cp src:resources uberjar $(UBERJAR)

FYI the command with no -cp before was including all the files in the folder where you were executing it

yeah, that's also not good.

this one now just includes the correct stuff in -cp

yeah

but this one works! bb -cp "src:resources" --uberjar $(UBERJAR) 🎉

cool! I logged the issue here: https://github.com/babashka/babashka/issues/776

PR ready for the former issue https://github.com/babashka/babashka/pull/777