LambdaMOO was(is) like this. And even more literally than in Smalltalk. Persistent objects, object oriented programming language, live code editing, object manipulation, programming right in the world, with immediate and shared effect.
But unlike a Smalltalk environment, the things you were editing were (in metaphor at least) living narrative "world" objects like rooms, objects, characters, etc. meant to give a (text-based) "VR" aspect. And it was multiuser, so collaborative.
It's a compelling, though exotic, way of making things.
If you'd like to literally live, breathe and walk around inside of your programming environment with your colleagues, set up your own https://dynamicland.org/ :D
Every time I look at one of their demos, I want to cry and say "I want this to be real". And then years pass, and there's another demo, and I'm re-reminded. This is more than vaporware, to be sure. But it's still nothing I can use in my day-to-day. Yes, it's research. It still makes me sad to see a possible future but no clear idea as to when it could become real.
So... I just did a little reading. And, I get the impression that if you want to do what I was suggesting, you'd have to physically walk up and talk to Brett Victor about it. https://dynamicland.org/2024/Is_Realtalk_open_source/ The entire goal of the project is in-person physical interaction.
At a glance I'm not totally sure I see if this is qualitatively different from "living in the shell." I definitely find myself spinning up new tools constantly, but they're usually in shell or Python, not Smalltalk.
The open nature of the environment led to data loss and corruption: crashes or forced restarts corrupted the Squeak image, faulty email synchronization scripts deleted server data, ad-hoc scripts sometimes created incomplete Person instances. The authors identified the need for transactions, backups, and crash recovery features. Storing all objects in memory limited scalability. And there is no mention of "performance" in the whole paper.
Whatever the waxing and waning that Smalltalk has had over the years, it's going to wane big time unless it can incorporate AI coding into their environments. Coding without it is going to seem ancient, soon.
"Our observations are derived from an experienced programmer using a Squeak/Smalltalk image as the primary environment for carrying out everyday tasks over a period of eight months."
> experienced programmer
So not actually an example of "knowledge workers themselves should ideally be able to quickly adapt the software tools."
23 comments
[ 3.0 ms ] story [ 59.2 ms ] threadBut unlike a Smalltalk environment, the things you were editing were (in metaphor at least) living narrative "world" objects like rooms, objects, characters, etc. meant to give a (text-based) "VR" aspect. And it was multiuser, so collaborative.
It's a compelling, though exotic, way of making things.
I'm attempting to defibrilate the whole thing here: https://github.com/rdaum/moor
The open nature of the environment led to data loss and corruption: crashes or forced restarts corrupted the Squeak image, faulty email synchronization scripts deleted server data, ad-hoc scripts sometimes created incomplete Person instances. The authors identified the need for transactions, backups, and crash recovery features. Storing all objects in memory limited scalability. And there is no mention of "performance" in the whole paper.
If only there was some decades old persistence implementation...
https://gemtalksystems.com/small-business/gsdevkit/
Oh!
> experienced programmer
So not actually an example of "knowledge workers themselves should ideally be able to quickly adapt the software tools."