This is a sleeper option, but 'xonsh' as a Python superset is a very pleasant experience, especially when you just don't feel like learning another bespoke language for the complex stuff.
I once had to scrape the output out of something hacked together by a grad student 10 years before me, scrape yet another hacked together application, do some fancy numerics, and then plonk it into a database. I did it without tears!
> AI model providers have zero "moat", clients change them as they see fit.
That might be true in tech-savvy industries -- but in non-tech industries where the biggest software purchase might be the office suite or the ERP, inertia means the GSuite shops stick with Gemini, and the Exchange/Office 365 shops stick with Copilot.
i.e., welcome to Constitutional law, where divining the effect of a amendment written in 19th century prose lives and dies on jurisprudential philosophy, rather than decided case law. (Is that what you mean by the “(and common law?)” parenthetical?)
Fascinating to think about the logistics with avoiding ROM – model compatiblity must've been a PITA! I'm even thinking about how to identify models – I know there's a pre-Gestalt Toolbox routine that doesn't need any Toolboxes, I guess?
I would give my left leg to learn how the permissions system worked – do end users (and PHBs) get to edit the rules directly? I fully expect some HR ass to go:
> Also...these things tend to have fuckin terrible documentation. Good luck figuring any of this out. And you can't google it and your AI is just as lost as you
I convinced my boss to hire an intern for the summer to do this. They said: "wouldn't internship projects that involved actual coding be more attractive?"
I replied: "Well, they'll be having to do a lot of experimenting to figure things out..."
> I do not like the R language at all myself, but to be fair there are reasons it is widely used in higher ed.
In the same boat... from a PL perspective, yikes (especially the macro mechanism that somehow never seemed to be planned, but somehow exists). As a working statistician? It really does get work done quickly.
To pass inputs with complex unevaluated syntax, I've seen...
– ad-hoc string parsing (lavaan etc.)
– formulas (which somehow the tidyverse doesn't use),
– base R syntax manipulation by round-tripping between as.list and as.call;
– and whatever wheel reinvention with bizarre semantics that the tidyverse uses.
Even though I've dealt with this, I'm genuinely appreciative of requirements: out of many stipulations, packages that monkeypatch are prohibited (I have a few ones that add diagnostics to advance analyses), online API access needs robust error handling... and there is a conformance/diagnostic suite.