[emacs-berlin] November Meetup next Wednesday

[Jack Rusher]

I was unfortunately unable to make it to the meetup, but I did check out Perry’s presentation.

One of the barriers emacs faces in remaining relevant is the long-standing requirement that everything that works on a graphical display also work on a TTY (for instance when running emacs on a remote server over SSH). I’ve been lobbying for years to eliminate this requirement so we can replace the display code (a hatefully complex pile of C code designed to handle low baud rate terminal connections) with something more flexible and graphical, but there has always been strong resistance.

In terms of the runtime, a rewrite in Rust seems relatively unpromising to me, and I feel that Perry overstates the benefits and understates the drawbacks of that effort. Instead, I feel we should throw more support behind the Guile branch. This work, which was basically all done by one programmer over a few Google Summers of Code, places emacs on top of the Guile runtime <https://en.wikipedia.org/wiki/GNU_Guile>. The Guile VM is fast, already supports elisp, has good concurrency primitives, and so on.

Ultimately, any effort to improve the runtime while preserving existing libraries is going to be fraught. It would be interesting to work on a “source-to-source plus human effort” translator to convert existing libraries into concurrency-aware code in a lexically scoped Scheme dialect, but it’s hard to imagine anyone supporting that work.


J.