Hey @U1M0BC7S7 - I noticed you liked that tweet and then I realized I should have taken it down. malli was for a brief moment part of nbb as an optional dependency, but I had troubles deploying both npm libs at the same time so I had to revert it unfortunately for now

A right, I thought it might be something like that 😄 Thank you for confirming @U04V15CAJ I'll happily try again at a more opportune moment.

@U1M0BC7S7 #babashka now does work with malli though - do you need malli specifically for something Node.js related?

I'm trying to convert an HTTP response checker from Clojure to nbb. I wrote it when I had no spec/malli knowledge and it runs in an always-on JVM process, so now I'm trying to convert it to a Serverless (AWS Lambda) model and also porting it to node/nbb for efficiency. I've done the footwork in #babashka but my skills for getting babashka to work in Lambda is not all there yet.

Ah right

For bb there is blambda and holy lambda

https://github.com/FieryCod/holy-lambda https://github.com/jmglov/blambda

😯 Ooh, I'm going to have a look at those. Thank you!

Yuuuuusss! got blambda going. Thanks for the pointer!

Nice! @U054AT6KT will be happy to hear this

What I've done a little differently is to create the resources via a CloudFormation template, which removes the need to keep track of the layer, role and policy ARNs while wiring it all up. Maybe not everyone's cup of tea, but it does make wiring up and tearing down the resources much easier 😅 I need some 💤 Thanks again for your help 🙏

@U1M0BC7S7 Cool. It'd be cool if you shared your findings somewhere sometime and then we could link to it from https://github.com/babashka/babashka/wiki/AWS!

@U04V15CAJ sure 😄 It's still a WIP, busy refactoring and clearing up things for a proposal: https://github.com/johanmynhardt/blambda/tree/cloudformation-poc/cfn In hindsight this looks like a lot more work, but once cleaned up I figure it would provide sleek layer + function management if abstracted away behind tasks. Need to get to actual work for today now.

@U1M0BC7S7 This is exactly what I had in mind. The lambda I'm working on uses Terraform instead of CloudFormation, but same basic idea. bb deploy was really only meant for getting started quickly. I’ll add TF and CF examples to the docs when I have time. 🙂

@U054AT6KT I stumbled across your Dogfooding Blambda! : revenge of the pod people post 😃 I tried to zip up a pod layer's content along with the runtime, and the bootstrap script has the BABASHKA_PODS_DIR set. In addition, the environment definitely has it set to /opt/.babashka/pods. Somehow I still get the error from your post where it's looking in some home folder.

:cause /home/sbx_user1051/.babashka/pods/repository/org.babashka/aws/0.1.2/manifest.edn (No such file or directory)

I seem to remember @U04V15CAJ saying that there was a bug with BABASHKA_PODS_DIR in some version of bb. Maybe try upgrading to the latest version, if it's not already?

yeah, you can use the latest master version from here: https://github.com/babashka/babashka-dev-builds

🙏 Thank you, going to try that now.

Yuss!

0.8.158-SNAPSHOT is 👌

@U1M0BC7S7 If you want to open a PR for Blambda to use this version, I’ll happily merge it. 🙂

OK, I'll have to add a change to support snapshot installs, the release URL is a bit different, I manually pulled and repackaged 😅 Can only have a look again tonight.

No hurry at all, and this is definitely an “only if you want to” sorta thing. 😉

There will probably be another bb release in a week or two

😄 I do want to. My OSS contributions are few and I'd like to change that.

Marketing took over on the No compromise headline

Performance is always complicated to gauge

I thought I'd be the pedantic commenter before anyone else did it ;)

Really nice article, thanks for writing it (and publishing during your vacation)

Where does it say sub millisecond startup time? I can't find it and don't remember writing that

> • Blazingly fast starts: sub-millisecond. AWS provisioning latency is your only overhead, so no worse than any other language and better than many.

🙋 The "no compromises" line was my fault Michiel. Writing pithy and intriguing blog subtitles is hard 😄

No worries, just wanted to take away that expectation before people came complaining to me ;)

"I had to compromise dammit!" 🔥

It does make for a good headline on HN

plz upvote! https://hn.algolia.com/?dateRange=pastWeek&page=0&prefix=false&query=clojure%20aws&sort=byPopularity&type=story

To be specific, the error that happened was:

npm ERR! code E404
npm ERR! 404 Not Found - PUT  - Not found
npm ERR! 404 
npm ERR! 404  '@babashka/[email protected]' is not in the npm registry.
npm ERR! 404 You should bug the author to publish it (or use the name yourself!

If the artifact was published correctly but the npm registry is responding incorrectly, maybe https://www.npmjs.com/support can handle a bug report?

I mean, the library was published before but updating it to a newer version failed

An alternative to this would be to upload the artifacts to Github releases and let nbb pull them from there

Since they are standalone, it should not be more complicated than just downloading them to a predicable location

Not sure what version it stopped working at. The above link to 0.5.112 does show as published on https://www.npmjs.com/package/@babashka/nbb-prismatic-schema

hmm, it seems the deploy did work then, but the client maybe responded badly?

however, 114 seems to be missing here for malli: https://www.npmjs.com/package/@babashka/nbb-metosin-malli

So due to this mess, I gave up on it

As you can see here, nbb + metosin malli succeeded here: https://app.circleci.com/pipelines/github/babashka/nbb/710/workflows/352ed3f3-3269-4c9d-a84b-258acfdea90f/jobs/634 but then at the end, schema failed

oh, it seems the version was wrong too:

npm ERR! 404  '@babashka/[email protected]' is not in the npm registry.

If someone can experiment with a setup like this, please let me know and I'll revisit

Hmm, maybe uploading the files to github releases first is good, and then I can retry the publish locally as well

@U1M0BC7S7 ^ see above for some context

Without any more context, I don't know what's wrong. What does load-token look like?

like this:

(def load-token
  (let [token (-> os
                  .homedir
                  (str "/access-token.json")
                  fs/readFileSync
                  js/JSON.parse
                  (js->clj :keywordize-keys true)
                  :access_token)]
    (if token
      token
      (prn "Please get a new access-token to proceed"))))

ah, did you mean (defn [] load-token) perhaps?

else you shouldn't call it as a function

uhh, yes... thanks

hey, I have similar issue with a.replace is not a function. Consider following snippet:

(cond-> (fs/readFileSync "resources.edn")
  :always (clojure.string/replace "{{base-uri}}" ""))

you need to first call str on the result of readFileSync since that returns a buffer, unless you give a file encoding like "utf-8" to it

(fs/readFileSync "resources.edn" "utf-8")

yes, it works, thank you!

I have made this mistake myself often too ;)