« first day (1285 days earlier)      last day (1666 days later) » 

Jeff Zeitlin
Jeff Zeitlin
17:00
/me always thought that ⎕IO←1 made sense; normal people start counting with 1, not 0. 0 as the origin for array indexing was because of lazy compiler/interpreter writers not wanting to be bothered with the additional correction when they implemented arrays as pointers.
/me notes that the first language he ever encountered that used 0 origin was PL/M-86, which he encountered in the context of being a low-level (as in, manipulate the iron) language for a proprietary workstation based on the iapx86-series CPU. He encountered C after that, and saw that both languages internally handled arrays the same way - as offsets from the address of the beginning of the array.
 
3 hours later…
dzaima
dzaima
19:47
@JeffZeitlin "normal" people start counting with 1 because that's how some (relatively stupid) people decided to count when zero didn't even exist or something. If starting with 0 has benefits (of which there are many), why not switch?
 
1 hour later…
Marshall
Marshall
20:53
Self-hosted BQN on top of Javascript is working!
It's made it through three whole test cases already!
(It is maybe kind of slow.)
ngn
ngn
21:05
@xpqz it's on hn right now
dzaima
dzaima
21:43
@Marshall that was not a joke.. i'd assume it'd be pretty worthwhile compiling the bytecode instead of interpreting it (not quite as nice as java bytecode, but generating JS (or maybe even wasm) on-the-fly shouldn't be impossible, not sure whether that'd be simper than just adding more native builtins)
Marshall
Marshall
@dzaima The thing to do first is to fix the search functions, since they're still using linear search. I think once I've done some work on the runtime I'll try a Rust version, and compile that to Wasm if it goes well. After that, start compiling to Wasm directly.
Just finished bcases.bqn...
So that's like 4 seconds per byte of input. Could be improved I guess.
 
1 hour later…
dzaima
dzaima
23:04
massively cleaned up my phase 2 problem 9 solution, and made 2 dzaima/APL variations (which output the would-be arguments to ⌹ as there's no ⌹ in dzaima/APL) - one which generates the expected coefficient matrix, and one which generates a bigger one to avoid all loopiness:
https://dzaima.github.io/paste/#0ZZHBattAEIbv@xRzs4SVKH6AgPscpYettFsPXUlGu6oMxqW04LomWxzi0kLpJVAIvTaBkEsPfZLMk3RWG/tSgWBn@f@Zb/@pAGi94/@SPn8xtYXRXLpZ7pq8al6iUadu4UbCgIZ2EFbnj5ebEe0vaH9L@90I2O9/gFHa5bozRdtVeYuvZg4qaV9b0cf@FW22POAZHL/g6tVRKXqMIP53L/kwzrP@PyHWpVoomx1qWTVd7YSO1l7COPpPJgfjE9LBKeR8kGqcLvUKepzS9hr6QSqNiTKYNxYdNr…

« first day (1285 days earlier)      last day (1666 days later) »