4 ms·

I really like the in-function documentation and unit like testing in the where clause. > # this is true > ((1 / 3) * 3) == 1 This worries me. I know it is tr

I really like the in-function documentation and unit like testing in the where clause. > # this is true > ((1 / 3) * 3) == 1 This worries me. I know it is true mathamatically, but experience tells me trusting floats is a recipe for disaster. A approximate equal operator is safer. Floats are such a leaky abstraction it is in my opinion better students know this early on.

klyrs 5y ago
The article doesn't really say what's going on under the hood but I'd guess that division of integers makes rationals... > Pyret has numbers, because we believe an 8GB machine should not limit students to using just 32 bits. Emphasis mine. WTF, author. But the Java sample below saying "this is false" implies that floats are a problem that is being avoided. The only reasonable solution is rationals... right?
- adjkant 5y ago
  Yep, this is a project out of the same world as Racket which does a similar thing. If you go into the editor/REPL (https://code.pyret.org/editor https://code.pyret.org/editor) and enter 1/3, you'll see that it gets rendered as 0.3 repeating (with a bar, not some arbitrary rounding) or if you click on it, directly as 1/3.
recursive 5y ago
`1 / 3` doesn't produce a float in pyret.
agumonkey 5y ago
it's a lisp descendant it has, most probably, a full numerical tower
- skrishnamurthi 5y ago
  Partial, not as full as Scheme.
na85 5y ago
Many languages treat the construction 1/3 As a rational type, and do not convert to float unless you cast. Common Lisp comes immediately to mind. Evidently pyret is another.
- tromp 5y ago
  Haskell supports rationals with module Data.Ratio, but uses a slightly different syntax: 1 % 3
skrishnamurthi 5y ago
You should try it and see what it does when you force it to use floats. You may find it interesting in light of your "leaky abstraction" comment. Pyret is designed precisely to make that point clear.
renox 5y ago
Floats are really a poor representation for reals, float-intervals like Frink does are much cleaner but our FPU/SIMD sucks for these kind just think about changing the rounding very often..

klyrs 5y ago

The article doesn't really say what's going on under the hood but I'd guess that division of integers makes rationals... > Pyret has numbers, because we believe an 8GB machine should not limit students to using just 32 bits. Emphasis mine. WTF, author. But the Java sample below saying "this is false" implies that floats are a problem that is being avoided. The only reasonable solution is rationals... right?

adjkant 5y ago
Yep, this is a project out of the same world as Racket which does a similar thing. If you go into the editor/REPL (https://code.pyret.org/editor https://code.pyret.org/editor) and enter 1/3, you'll see that it gets rendered as 0.3 repeating (with a bar, not some arbitrary rounding) or if you click on it, directly as 1/3.

recursive 5y ago

`1 / 3` doesn't produce a float in pyret.

agumonkey 5y ago

it's a lisp descendant it has, most probably, a full numerical tower

skrishnamurthi 5y ago
Partial, not as full as Scheme.

na85 5y ago

Many languages treat the construction 1/3 As a rational type, and do not convert to float unless you cast. Common Lisp comes immediately to mind. Evidently pyret is another.

tromp 5y ago
Haskell supports rationals with module Data.Ratio, but uses a slightly different syntax: 1 % 3

skrishnamurthi 5y ago

You should try it and see what it does when you force it to use floats. You may find it interesting in light of your "leaky abstraction" comment. Pyret is designed precisely to make that point clear.

renox 5y ago

Floats are really a poor representation for reals, float-intervals like Frink does are much cleaner but our FPU/SIMD sucks for these kind just think about changing the rounding very often..