diff options
author | Adrian Kummerlaender | 2019-04-16 00:12:37 +0200 |
---|---|---|
committer | Adrian Kummerlaender | 2019-04-16 00:12:37 +0200 |
commit | 908d8f01d5e964971ad76909c0ec31468ee93d2d (patch) | |
tree | f1f9a84928b39aabc4792544951797b8a9eb1bfa /library | |
parent | 1b92af67088b5e57f9134703ae6115c3529fb352 (diff) | |
download | slang-908d8f01d5e964971ad76909c0ec31468ee93d2d.tar slang-908d8f01d5e964971ad76909c0ec31468ee93d2d.tar.gz slang-908d8f01d5e964971ad76909c0ec31468ee93d2d.tar.bz2 slang-908d8f01d5e964971ad76909c0ec31468ee93d2d.tar.lz slang-908d8f01d5e964971ad76909c0ec31468ee93d2d.tar.xz slang-908d8f01d5e964971ad76909c0ec31468ee93d2d.tar.zst slang-908d8f01d5e964971ad76909c0ec31468ee93d2d.zip |
Using _vectors_ as fundamental datatype could make for a really neat experience.
Imagine e.g.:
* fundamental arithmetic operations that apply to both vectors and scalars
* implicit component-wise or vector-wise operations
* scalar values as 1D vectors
* higher order functions to manipulate those vectors
* `map` function that applies a (quoted?) word to all vector elements and returns the result
* efficient parallel operations
* a rich library of vector manipulation functions
* or even: matrices as a fundamental datatype?
* problem: probably harder to conveniently enter via a 1-D repl
* sound more and more like a RPN version of APL…
Back to reality, here is what this prototype actually adds:
* new `DList!int` based datatype alongside the existing int, string and bool types
* basic support for printing such values in a readable fashion
* new fundamental `:` word that constructs a vector of two elements
* `1 2 :` yields `[1, 2]`
* adapted `+` and `*` to support component-wise operations
* `1 2 : 3 +` yields `[4, 5]`
Diffstat (limited to 'library')
0 files changed, 0 insertions, 0 deletions