use https://github.com/clojure/tools.logging

Yes, eventually java libraries will log in a way you need to handle. This was helpful for me: https://lambdaisland.com/blog/2020-06-12-logging-in-clojure-making-sense-of-the-mess

In my opinion, logging is a chore and keep it as familiar as possible to new comers from Java. I use log4j2 with log4j2.xml to give config. There is no point in trying to have idiomatic clojure config. Log configuration is a one time activity. Keep it in a familiar xml format

OTOH java logging’s legacy’s such a pain. Not to say that today’s apps would better be served by something which combines logging/tracing more than traditional logging, which came from another development landscape, made of closed-walls enterprise systems

We like log4j2 + clojure.tools.logging (as noted briefly in that article) because you can avoid XML for configuration and easily provide dev/test overrides via env vars and system properties. It can also watch and reload the config when it changes (we use this in dev but not production). One day I swear I'll write this up in a blog post, building on the Lambda Island post! 🙂

can try https://github.com/lambdaisland/deep-diff2

Thank you

There's clojure.data/diff built in

• Anything you can do in a file, you can do in a REPL • The whole language is designed to make the REPL great.

The first page of the official REPL guide explains things better than I can :) https://clojure.org/guides/repl/introduction

For me, the most important effect of the clojure REPL on my workflow is the ability to get instant feedback on basically anything. That makes my learning loop faster than in other languages.

What "others" are you thinking of, specifically?

Plus one to "the whole language is designed to make the REPL great". This includes the way namespaces and vars work, to the focus on data and pure functions. The further you get from data + fns the more painful everything gets. This, to me, includes things that are irreplaceable in other languages, like dependency injection containers - if you feel like you need one, you may have strayed too far from the way of the REPL (end of controversial opinion ;) )

I was discussing about this with you previously and I think that the main difference is that Clojure makes it easy to reload parts of your codebase, including single vars (you have to delete the module cache in nodejs for instance when you reload a module). That being said, I tend to prefer the compile + watch + restart whole app workflow that usual node tools tend to propose (comparing with node js/ts here). When the transpiling / app init is fast it is more convenient IMO (you cant forget reloading things on your repl)

@U030H8RLWUU did that answer your question?

Yes, it did thank you

@U0PUGPSFR I especially thought about Python or NodeJS ones

Check the shadow docs for how to import js code. From there it depends on what cljs react wrapper you're using.

I use rsync to linode vps for my personal stuff

AWS CDK

Dokku on Digital Ocean and I even have a setup where I run everything outside of containers as well.

Dokku is primarily helpful for managing the lifecycle of your app in prod.

http://fly.io also supports deploying a Dockerfile (which you mentioned) and that is straightforward to setup.

Yeah pretty sure you can use Clojure an any of those platforms via Dockerfile. I looked at all three of those (and digitalocean's app platform) but ended up just sticking with droplets. I have a https://github.com/jacobobryant/biff/blob/master/example/setup.sh that I copy into all my projects, which (among other things) sets up git push-to-deploy. Easy enough once you get the script how you like it 🤷 @U6GNVEWQG I'm curious if you're able to connect to your production app via nrepl? I spent a day trying out dokku but could never get it to work, IIRC because dokku by default only allows HTTP connections to the container.

Deploying a Kit project to http://fly.io right now using the Dockerfile option. I've also deployed to fly using Jibbit to build an image and push it to their private registry.

I'm still unfamiliar with Docker and haven't liked using it on my own system. But in this case, are you literally just creating a Dockerfile in your repo and http://fly.io uses it internally? So I technically wouldn't need to use Docker during development, just create the file for to recognize and use?

It looks like 's free tier only gives you 256 mb of RAM. Is that enough to test out a hobby project? I guess it's only $6/month for 1gb of RAM but just curious. 's free tier is 512 mb (they were great when I used with rust apps) and they also accept Dockerfiles but their 1gb is $15/month

If you folks could share your dockerfiles I would greatly appreciate it but no worries if not

Yeah, I don't dev in Docker. I just deploy with it and that's only because it's "easiest".

I don't have a public repo I can share right now, but this is the template Dockerfile that Kit uses for new projects, which is basically what I'm using right now. You could adapt it for something else by replacing the RUN command with whatever you use to build an uberjar, and updating the COPY and ENTRYPOINT paths to reference the right jar file name/paths.

# syntax = docker/dockerfile:1.2
FROM clojure:openjdk-17 AS build

WORKDIR /
COPY . /

RUN clj -Sforce -T:build all

FROM azul/zulu-openjdk-alpine:17

COPY --from=build /target/<<name>>-standalone.jar /<<name>>/<<name>>-standalone.jar

EXPOSE $PORT

ENTRYPOINT exec java $JAVA_OPTS -jar /<<name>>/<<name>>-standalone.jar

Btw. AWS App Runner looks like an interesting option too: https://cloudonaut.io/fargate-vs-apprunner/ • notably, it supports https://cloudonaut.io/fargate-vs-apprunner/#:~:text=possible%20use%20cases.-,Scaling%20down%20to%200,-Our%20industry%20often.

Great info here, thanks so much everyone

Hey, apparently Railway supports Clojure without a Dockerfile as of today: https://railway.app/changelog/2022-07-22#clojure-langauge-support

Just to follow up, I was able to get a full stack clj/cljs app deployed on using a Dockerfile so it should be usable on all the other platforms we were discussing. It also has tailwind w/ hot code reloading too if others have been wanting to figure out how to do that. https://github.com/Chase-Lambert/clojure-app-template

Please note I don't know how to create an official template so I still have to manually edit all the TODO occurrences to match the name of the desired app/repo so if you know how to automate that, I'm all ears.

I tend to test the code my handlers call (and make an effort to keep the handlers "simple") and I test the API directly, using an HTTP client library (so I unit test the business logic and do end-to-end testing on the API itself).

I have a few tests directly against handler functions, but I don't use ring-mock for those -- I just pass in a hash map with the appropriate data.

thanks @U04V70XH6 👍

can you sketch out what you’re thinking?

I want a function that takes strings, splits on a generic regex pattern, and processes the substrings according to arbitrary logic. I was thinking the most idiomatic way to do that would be recursively, i.e. input is the contents of the Moby-Dick ebook, I split on \n, do some work, split on \s , do some more work, etc

But then I was thinking defmulti might also be an option?

Defmulti takes a single dispatch function. So is this a regex that matches a word and you have implementations for different words? Or do you envision running several regexes and then doing something on the first one that matches?

the former case

makes sense. I often find it helpful to run a bit of analysis over things to make a bit more structured data on top of strings

then you can dispatch on some token in your structured data and gives you a bit more chance to put more logic there rather than only dispatching on what a regex can do with a string

this is definitely just a toy project while I take a "tour" of clojure (which is a really cool language so far it seems) but that is essentially the idea. The ultimate project idea is to do ELisp-y text processing macros for editing, ie something like a macro that could remove repetitive word usages