> Beware, though it's fully functional, crosstalk is still limited by constraints of the original Smalltalk 80 system - e.g. in terms of color (black and white only), possible screen resolution (2^20 pixels, the system crashes if you try to increase the resolution beyond this) and available memory (~1 MB!).

I would imagine it would be less work to start with an early Squeak to remove these constraints than to try upgrading this system to remove these old limits?

Starting with Squeak is probably the easier way if you want advanced features such as color or network connectivity - still, there might be some interesting ideas to improve the system. One idea I had was to accelerate the bitblit functions using the Raspberry Pi's GPU. Currently, crosstalk uses a simple framebuffer and the runtime has to convert all monochrome bitblit operations into color ones since the Raspberry GPU doesn't have a 1bpp mode.

Support for multiple cores (available in all but the most simple Raspberry Pis) would also be nice. Here, the Roar system (https://github.com/smarr/RoarVM) might be an interesting starting point.

Let's see if I can find some students who want to work on some of these tasks... but my students are nowadays rather interested in Rust and RISC V :) (though a port of crosstalk to the Allwinner D1/Nezha and related boards and JH7110/Vision Five 2 is on my to-do list).