Welcome to Programmer Humor!
This is a place where you can post jokes, memes, humor, etc. related to programming!
For sharing awful code theres also Programming Horror.
(Please don't lob rocks at me. I love Python.)
To be fair, Python is just glue for code written in lower level languages when it comes to AI
A lot of it is c in a python raincoat
Which can be ASM in a C raincoat
Which can be ASMR depending on pronunciation and tone of voice.
depending on pronunciation and tone of voice.
Thank you I suppose?
The underlining linear algebra routines are written in… FORTRAN.
I've never played with FORTRAN, but I've done some linear algebra with matlab. Matlab was interesting for the native handling if matrices. What makes FORTRAN so good at linear algebra?
the main thing that makes fortran preferable to C is the way it handles arrays and vectors. due to different pointer semantics, they can be laid out more efficiently in memory, meaning less operations need to be done for a given calculation.
Interesting. Is this a fundamental limitation of C or is it just more preferable and easier to use FORTRAN when implementing it?
Meaning could the same performance be achieved in C but most optimized libraries are already written so why bother? Or basically C can't achieve the memory optimization at all?
you can get the same performance by using the restrict keyword in C.
basically, C allows pointer aliasing while fortran does not, which means C programs need to be able to handle cases when a value is accessed from multiple locations. fortran does not, so a lot of accesses can be optimized into immediates, or unrolled without guards.
restrict is a pinky-promise to the compiler that no overlapping takes place, e.g. that a value will only be accessed from one place. it's basically rust ownership semantics without enforcement.
Matlab's syntax for matrices actually derives from Fortran. There's a lot of flexibility in Fortran's array features for
Because Fortran does not have a pointer in the sense of C, the Fortran compiler is free to make several optimization that a C compiler can't. Compiled Fortran is often faster than C code that does the same thing.
That reminds me, I had a ride share driver named Blas, and I had to giggle and tell them about it.
Does one even have to actually write Python code, except for frontends? I'd assume you just load the model, weights and maybe training data into pytorch/tensorflow.
Doesn't seem to be the case, some popular servers:
And then of course talking to these servers can be in any language that has a library for it or even just handles network requests, although Python is a nice choice. Possibly the process of training models is more heavy on the Python dependencies than inference is, haven't actually done anything with that though.
Python-wrapped C, for the most part.
There's also a whole lot that's just C/C++ exposing a Python interface, without any wrapping.
Thats okay I love them both. But not equally. (Don't tell Python).
It sure made sense forty years ago. And I'd bet that the examples in that book are more AI than today's LLMs.
The dominant approach at the time were Expert Systems. This used a lot of carefully crafted data and manually curated facts that the inference engine can use. It also fit in a MUCH smaller footprint compared to conventional neural networks. But you also don't get real language processing, reasoning beyond the target problem domain, and stuff like that - it's laser focused and built on very small amounts of data. Much of the research from back then centers on using Lisp and Prolog of all things, so BASIC isn't a big stretch.
I am lobbing rocks at you because of that admission.
Ikr. Python does kinda suck as a language.
Every old timer knows AI is supposed to be written in Prolog.
One of the guys who taught me Prolog wrote the book: https://www.inf.fu-berlin.de/lehre/SS09/KI/folien/merritt.pdf
That book opening image is indeed telling
But there is one other, probably even more important advantage: Prolog is a programmer's and software engineer's dream. It is compact, highly readable, and arguably the "most structured" language of them all. Not only has it done away with virtually all control flow statements, but even explicit variable assignment too! These virtues are certainly reason enough to base not only systems but textbooks on this language.
The 90s certainly were a different time...
Love a language that doesn't care if you're using inputs to get outputs or using outputs to get inputs
I have this one! Probably at my folks' place, definitely I'll put it behind my chair so people can see it during video calls.
don't blame me i voted for turbo pascal
I loved turbo pascal. Anyone using that FOSS delphi equivalent thing?
Python hatched out of the egg on the cover.
Is Python not considered to be any good?
...It's okay. I've programmed in far far worse languages. ...It's got its advantages. It's got it's problems. 🤷🏻♀️
Edit: If you need a serious answer: Much like BASIC, it's a language often used in teaching programming. In that sense, I guess it's much better than BASIC. You can, like, actually use it on real world applications. If you're using BASIC for real world applications in this day and age something has gone really wrong.
Python is great, but it's so forgiving that it's easy to write garbage code if you're not very proficient and don't use the right tools with it.
The only objectively bad (major) thing against it is speed. Not that it matters much for most applications though, especially considering that most number crunching tasks will use libraries that have critical path written in a systems language:
numpy, pandas, polars, scikit-learn, pytorch, tf, spacy; all of them use another language to do the cpu intensive tasks, so it really doesn't matter much that you're using python at the surface.
It's certainly not very fast
Python itself might not be, but all the AI shit runs on GPUs so it's CUDA or OpenCL or whatever underneath
Python is phenomenal for prototyping IMO.
Once you need performance, its best to use another language (even partially).
But quickly banging out a concept, to me, is the big win for python.
Once you need performance
If you need more performance. Many things just don't.
But quickly banging out a concept, to me, is the big win for python.
For me the best language for quickly banging out a concept has always been the one I'm most familiar with at the moment.
Python is the tradeoff between ease of development and performance. If you do things the "normal" way (aka no cython) your programs will oftentimes severely underperform when compared with something written in a relatively lower-level language. Even Java outperforms it.
But, you can shit out a program in no time. Or so I've been told. Python is pretty far from the things I'm interested in programming so I haven't touched it much.
good is subjective to the opinions of the group.
objectively, Python is a smoldering pile of trash waiting to completely ignite. it does have one thing going for it though.
it's not JavaScript.
It's okay, but it's a bit slow and dynamic typing in general isn't that great IMO.
Dynamic typing is shit. But type annotation plus CI checkers can give you the same benefits in most cases.
One tremendous strength of Python no one has mentioned is its vast ecosystem of high quality packages. It's not just the language features that speed up development, that ecosystem makes a huge difference.
Another (far more subjective) advantage is readability - when written according to Python's (actually quite opinionated!) style guidelines and general software engineering best practices, Python is also extremely readable, which really facilitates teamwork. My software shop has transitioned to using Python for most things these days for that reason, away from JS, after seeing my work and code reviews, FWIW.
I'm not some wizardly dev, to be clear, but I'm this shop's first senior dev specializing in Python. I write deliberately clean and readable Python and folks are really enjoying it - enough to voluntarily switch.
Performance is always listed as a Python drawback, and it's not untrue, it's just so overblown as a problem. It basically never causes me issues. Crucially, saving dev time is almost always the better choice compared to saving compute cycles. And I'd take that farther and say anyone junior enough to be wondering about Python and performance...is almost certainly working on tasks that Python is well suited to - better suited, than most other languages.
(Hopefully this was not too controversial, but I accept the risk of a flame war, as is tradition lol)
Edit: clarity
Very good explanation, thank you.
Cheers friend!
As far as I know many Python libraries which need performance are mainly written in C++
Would it have been any less shitty if it had instead been written in assembly?