💾 Archived View for republic.circumlunar.space › users › dbane › gemlog › 2024-03-31.gmi captured on 2024-08-18 at 17:39:40. Gemini links have been rewritten to link to archived content

-=-=-=-=-=-=-

Long-Term Trends

My Gemlog

Looking over the entries in this gemlog I notice that my interests stay constant for a particular era, then change to a new language/technology/paradigm. The overall aim is to try make programming simpler/higher quality by using (often old) research that industry has bypassed for whatever reason (usually the non-technical reasons of marketing budget & backing by a multinational company).

The ISLisp/EISL era. This was great, I learned a lot (about C as well as Lisp) but in the end the quality of the Lisp ecosystem looked like turning a prototype into a product would fall short of what I was aiming for.
The safe imperative/OO - Oberon and D - era. This started as an adjunct to EISL, the long-term benefit for me may be that when I see a new similar paradigm language like Rust I can ask "what does this offer me more than Oberon" before spending time on it.
The Standard ML era. This ended by changing an opinion I held for a long time, unfortunately I no longer think ISO or similar standardized languages are adequate.
The Concurrent Logic Programming era. Started with KLIC but then switched to FLENG which has been much smoother for me.

There are also some individual posts, e.g. about document preparation or audio programming, but that's mostly it.

The Craft of Programming

My hope with the above was to make it easier to reason about algorithms, e.g. the ideal for logic programming is that you can define a program using statements in FOPC. But for the past 10-20 years there is a separate trend in industry where you just code "glue" between components in some package repo like npm for JavaScript; pip for Python, etc. Programming language capability doesn't matter as much for this--it seems to be more network effects & multinational company backing (or at least tolerance).

But a computer security attack using the "xz" compressor has been all over the news lately, this *should* lead to an increased emphasis on supply-chain security and reverse this trend somewhat such that you use much fewer components from entities you trust, and verify. So perhaps learning different paradigms may prove useful after all, although I was really only motivated by curiousity.

Back to my gemlog