(he/him 🏳️‍🌈)

Computers are bad

Web: http://patpatpat.xyz
What: Teaches maths/computer science at UoL
Where: Leeds, UK

Last Notes

- reply

I had the same feeling recently (been learning Forth and using a lot of Scheme)
Cons = Push
Cdr = Pop
Quote and unquote = [ and ]
Functions are lists = Words are stacks
(Potentially heretical, but) parentheses are fiddly = Stack manipulation is fiddly
(I'm sure there are many other equivalences!)

2025-11-13 17:04:29 UTC

Latest ThinkPad update (I should probably name her?)
https://mastodon.sdf.org/system/media_attachments/files/115/543/467/383/963/220/original/d75d8c901fbd5844.jpg

2025-11-09 14:36:42 UTC

@npub1z2c…laft you might enjoy this cute bookmark I just got (from a bookshop on a boat in Leeds, called Holdfast)
(I decoded it with Rabbit Waves. I love the note about Japanese semaphore, I always enjoy learning about non-Latin encodings to stuff)
https://mastodon.sdf.org/system/media_attachments/files/115/520/234/955/441/463/original/fda33c3693b6f443.jpg

2025-10-16 19:30:28 UTC

- reply

(Not really relevant to this thread as I used Scheme, so feel free to disregard!)
I did it. It works (if by "works" you mean a relative error of less than 0.1)
I also stumbled across Stirling's Approximation, which actually can be derived using Poisson and Normal distributions! So my silly idea was not completely nuts!
For the Poisson bit, I used Knuth's algorithm to generate random Poisson-distributed numbers, and used that to find P(X = k)
https://patpatpat.xyz/data/scheme/badfactorial.scm

2025-10-15 19:05:07 UTC

- reply

this gave me a very stupid idea: the PMF of the Poisson distribution uses a factorial, therefore you could (if you were feeling suitably whimsical) calculate factorials using it
i.e., k! = mean^k / (e^mean * P(X = k))
Where P(X = k) can be found using a Poisson distribution table of your choice
(Needless to say, the accuracy is atrocious)

2025-07-27 02:05:19 UTC

I wonder if there is an alife/automaton to simulate a crowd at a rave/gig
E.g.:
- Urge to fill space and move to front
- Constant oscillating movements in time with music
- Simultaneous events (cheering, arms up, etc)
- Some attractions between friends
- Some enclaves of friends with a weak circle-forming urge
- Ability for people to pass to exit, via the classic hand-on-shoulder method. Possibly forming a chain of people

2025-07-13 15:17:05 UTC

- reply

I always want there to be "one correct way" of doing things, but unfortunately that essentially never happens in computing (or anywhere, really)
I think my pure maths background leads me to want a direct indisputable line of reasoning, when in reality there are multiple paths and all of them are wrong/right (in slightly different ways)
So, yeah having flexibility to do both methods is probably the way to go: pragmatism over dogmatism

2025-07-13 08:48:42 UTC

Length-prefixed strings (length + string) are so much nicer than null-terminated (string + 0)
E.g, if I want to save my string to a binary file and read it again, I know immediately how much memory to allocate for it (rather than reading it once and backtracking)
Also removes all the unsafe str functions, by forcing the use of the strn functions as default
Perhaps a big issue is that a string is now sizeof(size_t) + sizeof(char*), rather than length + 1
I am open to counter arguments :)

2025-04-03 23:39:50 UTC

- reply

Ooh that is very cool, it really shows off how great stacks are as a construct. Its quite fun that a good way of performing infix maths is just to convert it to RPN
Mixed-fractions, decimal/fraction conversion, and some misc functions are on the TODO list in my head. I probably also need to add variables, as currently the result is just printed out?
I actually just treat everything as a fraction, so there's not really a concept of separate digits. Its all just reciprocals

2025-04-03 23:01:57 UTC

I've made a fraction-based calculator, for simplifying rational expressions (written in ANSI-C using vi)
I was aiming to merge a scientific calculator with "bc", because I often need to simplify various fractions and whatnot. Quite happy with the result!
Uses shunting yard algorithm to evaluate the expressions, and a naive descent parser (I think?) to tokenise
Aside: OpenBSD's "editline" is very easy to use
Please ask any questions :)
http://patpatpat.xyz/data/frac
#c #bsd #unix #math
https://mastodon.sdf.org/system/media_attachments/files/114/276/472/321/733/302/original/81f8c7475cc9ed07.png

2025-03-06 19:37:33 UTC

I should learn how to use mailing lists and IRC.
Because a.) they have cool people on them, and b.) they are so old that they are essentially immune to the enshittification wafting over the waking world.
My only issue with IRC is that it seems like you have to be constantly connected in order to actually keep track of a conversation? I like COMMODE on SDF because it has a nice room history (as well as a friendly TUI). Any tips are appreciated.
#irc #email #mailinglist

2025-01-01 20:54:15 UTC

For reasons unbeknownst to God, I have been writing a BASIC interpreter
This book has been very useful. Its a really clearly written explanation of the inner workings of the BASIC interpreter for the Atari 8-bit computers. Includes all the data structures, memory layouts, language syntax, etc
Interestingly, it calls it a "pre-compiled" interpreter because it converts the user input to numerical tokens. But I guess we would call it "bytecode" nowadays?
https://archive.org/details/ataribooks-the-atari-basic-source-book
#atari #BASIC
https://mastodon.sdf.org/system/media_attachments/files/113/755/086/678/802/616/original/0cce14ae63f99daa.png

2024-11-23 09:43:45 UTC

- reply

I find it quite depressing to imagine a future where teachers just use AI to create courses, and students just use AI to attend courses
It presents learning as a means-to-an-end, and attempts to skip the difficult bits (like having to read, understand nuance, search for sources, discuss with peers). There is no short-cut to learning, the difficulty IS the learning (a quote I like is, "memory is the residue of thought")