I miss Smalltak so much ! Was the first serious language I learnt, playing with it was incredible.

Mess with Pharo?

I messed with pharo in summer 2022 while on vacation.

It looks featureful on the outside but when you look a bit deeper it shows that either sone libraries are half-done or completely rotting inside.

Could you give some more concrete examples? I'm curious as to what you mean. I've worked with Pharo but not with other Smalltalks.

I remember trying to play with a web framework (z something?) but it felt half baked and the documentation was fairly lacking, for example.

In terms of documentation, several free books on Pharo are available at https://books.pharo.org/ As it's development is fast paced, they tend to get a bit outdated fast. But at least the basic ones get updated often. Most recent versions are from 2022. I think the framework you are referring to is Zinc.

You might mean Zinc? Actually it is a thoroughly modern http client/server library. It is documented here: https://books.pharo.org/booklet-Zinc/.

I based my own web framework on it. It is very featureful and well-written.

You can check:

https://github.com/zeroflag/Teapot

https://github.com/demarey/pharo-rest-tutorial

The readmes should give you a nice idea

Cuis Smalltalk fixes that problem (at the expense of having fewer libraries):

https://cuis.st/

I am not sure if I’d prefer fewer libraries (of alleged higher quality) or more libraries in a rotting state.

To be completely honesty if I had to pick a smalltalk to start messing again I’d pick pharo again to go with the most popular implementation.

I really appreciate the design principles behind Cuis. Great to see it mentioned here.

I'm sure there must be plenty of people (like me) who have heard good things about Smalltalk over the years but haven't had much chance to explore it. If you want to dip your toes in the water, and if you agree with the following quote from Cuis's lead developer (Juan Vuletich), Cuis is definitely worth a look.

"Unbound complexity growth, together with development strategies focused only in the short term, are the worst long term enemies of all software systems. As systems grow older, they usually become more complex. New features are added as layers on top of whatever is below, sometimes without really understanding it, and almost always without modifying it. Complexity and size grow without control. Evolution slows down. Understanding the system becomes harder every day. Bugs are harder to fix. Codebases become huge for no clear reason. At some point, the system can’t evolve anymore and becomes “legacy code”.

Complexity puts a limit to the level of understanding of the system a person might reach, and therefore limits the things that can be done with it. Dan Ingalls says all this in “Design Principles Behind Smalltalk”. Even if you have already done so, please go and read it again!" [0]

[0] https://cuis.st/features

The Cuis developers are working on a great introductory book that I’m working through: https://cuis-smalltalk.github.io/TheCuisBook/

But that can happen with any ecosystem, right?

It is expected that you will use only the solid ones for production in any tech stack.

If you're willing to boot to Windows, Dolphin Smalltalk is the best modern day version. It's great fun to play with and easy to make native tools.