Time to first plot? We need a simple egg timer graphic. Or maybe an animated paperclip. Clippy must be looking for a job these days…
8 Likes
Indeed. That reminds me that Julie 2.0 should save the code into some closed format binary file.
2 Likes
Julie 2.0 should make the Rackauckasverse into a stdlib.
7 Likes
Clippy is fine. He has a home in kakoune these days
1 Like
To give extra time to Clippy (or the fantastic new logo) and save Stefan’s eyes, Julie 2.0 should automatically load all registered packages. No more using or import (or from)! Any function you could need is already ready to go. If names conflict, one will be chosen randomly; they probably do about the same thing anyway.
11 Likes
“I see you’re trying to write a molecular dynamics simulation of the elasticity of blood vessels in collagen deficiency… would you like help with that?”
6 Likes
We need more verses: Timholynessverse, Davidanthoffverse and (very important cause PERFORMANCE!!!) Chriselrodmagicverse. How about the Tamaspappgivingheartsverse and the Cormullionbeautyverse? Definitely needed! For general inclusiveness I want to have the Chakravalaingeniousbutdifficultverse!
More cowbells!
The only aim can be completeness! And it must be Julie 0.1, the zero/one and only.
11 Likes
Write an OS in Julie 2.0 to solve the two-language problem of Linux (C & Bash). The interactive shell will run the same language as the OS kernel. Security updates will be delivered by the package Revise.julie.
6 Likes
So we should call it VerseVerse? Base.VerseVerse.Batteries.ChrisElrodMagic.VectorizedRandomness.fast_rand!(x). This has more dots and therefore it’s better than what we have right now given what I hear on Twitter.
12 Likes
I find C++ namespaces to be easier, we should adopt them.
std::Base::VerseVerse::Batteries::ChrisElrodMagic::VectorizedRandomness::fast_rand!(x) is more readable for me
8 Likes
Whatever happens with Julie 2.0, I hope JuliX 3.0 will switch to directly writing IR on LLVM. All this syntactic sugar on top of that is harmful.
I also suggest picking up versions left behind. Julia 0.8 and 0.9 were unfairly excluded from being released. Julie 2.0 should be 0.8 and JuliX 3.0 should be 0.9. #NoVersionLeftBehind
7 Likes
Can you change array indices to use the machine memory address? All that abstraction just obscures the code.
7 Likes
I wrote self-modifying code in a computer science course at University.
Julie definitely needs self-modifying code.
Computer science experts - is it the concept of being able to self-modify which sets Turing machines apart from simple mechanical machines ?
1 Like
Julie targets a custom machine architecture for hand-tuned performance:
9 Likes
For those of you not on the v2.0 triage calls, Julie 2.0 has been renamed after hours, days, and weeks of bikeshedding. The only two contenders were Julia 2.0 and Julie 2.0. On one hand, the pro-Julia discussants had rational arguments wrt the burden of changing the spelling, but on the other hand, the pro-Julie discussants were very loud and kept arguing and arguing to no end. So the best solution is obviously to let both sides have a say, and the compromise is Juliæ 2.0. To accommodate the annoying use of æ, NumFocus is sending out Danish keyboards to the first 100 Juliæ users to download Juliæ 2.0. They are unfortunately sourced from cleanup in an old IT department so you will need to buy your own PS2-to-USB converters.
39 Likes
So long as Julie 2.0 has diagonal major matrix storage, I’m in. Python uses row major, FORTRAN uses column major. So diagonal major would be exactly the right compromise. (And why bother with off-diagonal terms anyway? – their sole purpose is to make matrices difficult to invert.) Same applies to tensors.
25 Likes
Will Star Wars indexing finally become the default?
12 Likes
Is that where 4,5,6,1,2,3,7,8,9 is the order of storage, and it takes decades to index anything?
8 Likes
It is.
2 Likes
I prefer Microsoft Windows indexing: 1, 2, 3, NT 3.5, 95, NT 4.0, 98, 98 2nd ed, 2000, Me, XP, Vista, 7, 8, 8.1, 10, 11, ???
17 Likes