>>12780181>>12780150>>12780103>>12780084>>12780079>>12780071Not the anon you’re replying to, but here’s my take
Matlab is great as an extended calculator. I’ve written at least 10,000 lines of valuable computation in it, and simulink is ridiculously good. However, it has a host of problems.
>syntax It claims to be made so you can copy off of a math text or paper easily and do the numerics, but in practice it ends up being a horrible mix of both conventional programming language syntax and poorly transcribed math into standard ASCII. There are often 5 ways of doing an elementary operation, all of which feel like bloat that ends up making matlab way more verbose than elegant. On top of that, it make’s it a pain to read other people’s work since I constantly have to deal with sloppy habits unenforced by a language that wants all syntax to be valid.
Also, I have a sneaking suspicion that matlab’s “lookup libraries and mash everything into the calculator” style probably bears the most weight for convincing engineering students that software related tasks are library plug n chug - it’s these student’s only experience with code!
>paywallIf I want to use my prior work in more general settings, or if I want to link my code in matlab to other languages, I have to lug around this 30+ GB expensive, proprietary library. I can’t run extra compute on my personal server box.
There’s this feeling that matlab persists because it’s what engineering students were trained on, and it’s convenient, and so they only really know matlab for numerics. I can’t meaningfully have my work shared among scientists and engineers on, say, Github where I know they could work on it with their personal machines. I can’t publish research code that doesn’t take this expensive software instrument. That might be fine for solidworks folks, but I refuse to accept that as anything but a big negative.
