http://neopythonic.blogspot.com/2009/04/tail-recursion-elimination.html?m=1 GVR has been on record for a while as disparaging of functional constructs, so it makes sense that doing things the "Pythonic" way means rejecting them.

brb redoing all of "Data Science from Scratch" in Hy

I actually called Clojure from Kotlin it's easy, not sure about the other way around, I guess you have the same kind of problems as using Kotlin from Java, https://kotlinlang.org/docs/reference/java-to-kotlin-interop.html

has electronics or hardware experience helped you in the IoT? hacking on arduino or raspberry pi?

Nope. Being able to understand basic binary operations mostly. We work with a variety of IoT devices, each of them talk to us over various networks, so decoding packet data (from specifications) is all you need mostly

big endian/little endian work helps too 🙂

is the JVM helping by virtualizing your code on the different devices?

nope

we have hundreds of microservices consuming from data being ingested via streams from these IoT devices.

those microservices are written in kotlin

Ok, then you must be using Spark or Storm?

Nope

Just plain TCP network data, into Kafka 🙂

you just collect the real time data directly into Kotlin?

no message broker que?

kafka

ahh ok, apache product… nice

I am getting ready to build a real time data pipeline from a chat system

We have had great success with kafka, however, YMMV.

I can’t say I know enough… to compare Kafka, Storm and Spark

actually, I first started by looking at Golang but then changed my mind about it for this use case

Start simple, try kafka, then determine based upon your observed outcomes whether you need to add in more pieces, like storm/spark and all that entails.

ok… sounds good. I will try Kafka for the first data pipeline

If you can get messages streaming into kafka, and your downstream consumers reading those messages in a scaleable way, then you determine whether it's fit-for-purpose 🙂

gotta go, home time.

ttfn!

ttyl

btw, clojure has nice libraries for working with kafka (like jackdaw)

ya, I probably can use Clojure for the data side. And Kotlin for the web stuff.

you might want to try an experiment of doing a parallel coding exercise between clojure and kotlin for the same problem and see which one you like

k

ok, I will do that

byebye

do you have any ideas for testing the trade offs?

Informally, I’d aver that clojure is aimed squarely at folk implementing systems that manage data, with good stories to tell around handling concurrency. It could be said to trade away some performance in that pursuit, and has a higher learning curve for most folk. The story I’ve heard around kotlin is mostly “java but better” but I suspect that’s giving it short shrift, so I’d be curious to hear the perspective of folk actually using it, esp. if they also use clojure.

I use Clojure for scraping web pages. I like it because it is succint. I am very new to Clojure though. I know it shines in the data world.

I think Kotlin comes from a very pragmatic "middle-way" approach that doesn't go to extremes (such as Scala) or verboseness (Java). It was designed to get the job done, easily, succiently and with the mimimal of learning curve for java programmers

We find it easy to write very terse, easy-to-follow programs that are also easy to test and maintain and (re)training java developers to use kotlin was very painless. Everyone had abandoned java (given the choice) after a couple of days of working on kotlin

(with some hand-holding for a while, until they got better and better)

We use everything in the section What Kotlin has that Java does not:

https://kotlinlang.org/docs/reference/comparison-to-java.html

Liberies like Lombok offer some of the same things to Java, but because it's not first class it's not working as well. While Kotlin is a bit more functional then Java it's way more oo then Clojure. I would choose Clojure over Kotlin myself. But if I would choose for a team of Java developers it would probably be better to go with Kotlin

Ya, I still love Clojure. I do like a few things about Kotlin… built in the null object pattern, composition over inheritance, async but with procedural style code (no callback hell)

About the null safety, I had the same implementation in Clojure and Kotlin. But when switching to Kotlin the program crashed because there where null's in the database. Some issue arises with interop. In that respect I much rather like the Clojure approach of almost always returning nil when called on null/nil. But not always since (read-string nil) does throw an error. Seems like only an advanced type system like with Haskell or Rust can really safe you here.

I wrote a null object pattern in Ruby. I wonder if I will need that pattern in Clojure. It’s usually useful in places like a view template… so an error isn’t thrown when data is nil/null

First make sure all changes also are on remote branches. Then you could either rebase or merge (locally). Once your sure it's working and contains both changes push.

https://stackoverflow.com/questions/449541/how-to-selectively-merge-or-pick-changes-from-another-branch-in-git just found this, which also suggests a rebase

git cherry pick might help too. it helps you only merge specific files, 1 at a time.

I’d make sure to create a new branch just to do the rebase or cherry pick… in case it goes wrong

Ah yeah, was thinking about git cherry pick. Unfortunately there’s a bunch of commits but only some have files I want. But yes, I’ll create a copy of the branch so I can try out these methods first.

As long as the branch your rebasing already exists remote, and you only push after your confident it works you should be safe.

pretty much, the only rebase that I use quite often is that

git squash

I rebase all the time

I just have some pristine copy of my code somewhere on my FS

just in case :’)

I rebase as much as my team members allow me to, which basically means just the feature branches only I work on.

yeah that makes sense. Also, rebasing isn’t perfect for “catching up”.

I really would like to rebase more but that's only possible if we would all work on our own forks, an then rebase on the origin's fork.

wasnt it you @U09LZR36F ,that linked the git flow used by git itself?

Ah yeah, here: https://hackernoon.com/how-the-creators-of-git-do-branches-e6fcc57270fb

Though I may be able to take that updated, rebased old and just merge it into new since it would only have the files I care about, and be up to date with the latest stable at that point.

Anyway, thanks everyone for the help\suggestions so far. At least now I have a plan 🙂