Same here... back in 2011 or so. We needed something much more performant than WURFL. My efforts eventually became a feature/product at Akamai known as "Edge Device Characterization" (EDC) using algorithms not dissimilar to how LLMs are trained today.
I can't speak to how good the actual product is today (or even when it launched, but that's a whole 'nother story), but during development it was capable of processing 100K RPS in a footprint of ~30MB RAM with ~98% accuracy compared to WURFL as a baseline.
I've had great success with a parser-generator known as "Marpa" [1]. One of the things I particularly liked about it is how simple it made to emit accurate and useful error messages when parsing complex languages. In addition, it can handle anything that can be expressed in BNF, and it's quite fast.
I agree, as I've missed a dose here and there over the years and experienced the intense paresthesia and headaches and nausea from withdrawal.
Background: I've lived my entire adult life (and most of my teens) with severe chronic depression. In my early 20's I started taking pharmaceutical treatment, and once I found the right drug (after trying many over the course of years) my life became manageable. SSRIs helped but I experienced severe nausea on most of them, or worse. It was only when I tried SNRIs like Effexor that things started to get better. YMMV, IANAPsychiatrist, etc, etc.
A few years ago I switched from Effexor to Cymbalta. Same class of drug - The Effexor simply wasn't helping as much as it used to and the switchover was done with a long taper-down and replace period. I even bought a lab-grade scale to measure out the contents of the capsules so I could cross-over smoothly.
All that said, Cymbalta has the same withdrawal effects, on about the same time scale - a single missed dose. But I wouldn't give it up unless something better comes along. I still struggle with my depression and the SNRI is just one tool in my toolbox for managing it.
It's a pretty impressive process. The cable is indeed hoisted up, after being "hooked" by a device dragging along the ocean floor. I think I saw a video of this once, here's one on youtube (not the same one I saw, but seems to have the same info): https://youtu.be/m6qTk5WNq9E
Smalltalk is definitely underrated. It can be an extremely productive language, and has built-in capabilities you won't find in any "mainstream" language that make coding in it very pleasant. :)
This depends on your definition of "underrated" but I'd definitely throw Perl 5 into the ring here.
Edit: Specifically, Perl 5 + CPAN is what makes it so much better than many people think. The language itself is insanely flexible, which lends itself to extensions that greatly increase its expressiveness. IMO, if you start with Moose (from the CPAN) as part of your "core library" Perl 5 becomes a very powerful tool for writing very nice code, at any scale.
I can't speak to how good the actual product is today (or even when it launched, but that's a whole 'nother story), but during development it was capable of processing 100K RPS in a footprint of ~30MB RAM with ~98% accuracy compared to WURFL as a baseline.