@borkdude Just checking how 0.5.1 is looking. On Wednesday I’m going away hiking for the rest of the week, and I have the Cursive bit in a state where I could release a beta build. Since it’ll require 0.5.1 for the deps output part, I’ll wait till that’s out before releasing. No pressure though, otherwise I can release when I’m back, no problem.

I’d like to use a glob in fs/modifield-sinceit would require another branch here: https://github.com/babashka/fs/blob/master/src/babashka/fs.cljc#L720 Is this worth a PR?

@donavan you can use it like this: (fs/modified-since ... (fs/glob ...))

In the anchor?

should also work I think

Don’t think so, it would mean at that line above you have a vector of file paths

…which ends up being passed to fs/file

ah ok, so the anchor is expected to be one target file

representing a file or directory, yeah

I guess for now you could use (mapcat #(fs/modified-since % ...) your-files) ?

true

but yeah, it would be nice if you could put a glob there

Shall I take a stab at PR?

I'd have to think if this would still be the correct semantics, but since you can put a dir there too, I think it would

ok, please add tests :)

will do 👍

I will publish a new bb tomorrow to support new efforts in #cursive and LSP to get better babashka script support (resolving of vars for static analysis + nREPL). I upgraded deps.clj to match the newest clojure CLI (so bb clojure should work with the new tools stuff as well). Please give it a spin. Linux: https://21422-201467090-gh.circle-artifacts.com/0/release/babashka-0.5.1-SNAPSHOT-linux-amd64-static.tar.gz macOS: https://21427-201467090-gh.circle-artifacts.com/0/release/babashka-0.5.1-SNAPSHOT-macos-amd64.tar.gz Windows: (build still going: https://ci.appveyor.com/project/borkdude/babashka/builds/40221161/artifacts) /cc @cfleming @ericdallo

This may be a bit out there,,, but has anyone an example of using babashka to convert a series of clojure hash maps into individual curl scripts ? I want to populate a dynamodb database via an API I have for staging and production environments, which I'm supposed to use a script (although will anyone know if I switch on an nREPL server...)

How general are your maps? You may be able to use selmer to convert them into the text you want. The difficulty of that approach is correlated with how many curl features or flags you want to use

an example of one of those maps would help us understand your problem

Its a completely flat map with keywords and mostly strings as values {:account-id "" :account-name "" :status true ,,,} and about 6 other key values pairs the same

@jr0cket do you mean, you would like to insert this as JSON with a POST request and you would like the curl command line format for this?

Suppose I’ve written a babashka script with a couple of external project dependencies; what’s the best way of packaging that up into something a not-super-technical person without babashka installed could easily download and run?

I was thinking applying uberscript and https://github.com/MnRA/nativity might be the solution, but nativity doesn’t seem to work for me at all when playing with it locally

@aengelberg actually, the upcoming 0.5.1 version has a new CLI option calls print-deps which prints a deps.edn map with all built-in deps + external deps

you could then use that for making a native CLI. I've illustrated here how you could do that, once you have a deps.edn: https://github.com/borkdude/jayfu

it would be cool if you could test that new functionality

else wait until tomorrow

I can provide you a pre-release binary, just let me know

but to be honest, just letting the other person install babashka is maybe easier since you won't have to deal with all your consumer's OSes, etc.

you could package up the script + deps into an uberjar and let that person run that

I can have a hack around with something in the morning... I am exhausted now.. its been a long 12+ at work 😞

bb uberjar does that

interesting

btw thanks for all the work on babashka, just started using it recently and really enjoying it!

I’m not really sure how binaries work, but do you think it’s possible (for each OS) to combine the existing bb binary and a script into a single binary that just feeds the script to bb?

without really knowing the internals, it feels like that should be technically feasible without having to re-run GraalVM

this feature has been requested more often, but this is similar to packaging up a complete nodejs or python + scripts. Usually it's easier to just install node on the other side and them let them use the script

Perhaps it is possible to package some extra resources into a binary, but I don't know how

running a bb uberjar requires JVM installed, right?

no

only bb

ah

only when you download / compute the classpath it uses a JVM

but once you have created the uberjar, no JVM needed

also not for creating the uberjar itself when you already have the classpath

To be sure I asked the binary + resources question in the GraalVM slack (https://graalvm.slack.com/archives/CN9KSFB40/p1627930036001500)

thank you!