Skip to content

Latest commit

 

History

History
86 lines (65 loc) · 2.77 KB

0.6.0.md

File metadata and controls

86 lines (65 loc) · 2.77 KB
Raw

0.6.0

Performance.

This release is FAST. Very fast.

Random data generation is now 1200% ~ 35000% faster, and uses 1900% ~ 33000% less memory.

Obviously, unless you throw the random data you generate away without looking at it, you won't see a speedup in your tests of that magnitude. Antecdotally, in my work-project's tests where Nyaya is used alongside an equal amount of typical unit testing, total test times dropped between 50% to 80%'ish. That's a massive saving!

New Random Data Generator

Mostly for performance reasons, Nyaya no longer depends on NICTA/rng for random data generation.
It instead comes with its own, new module. (examples)
(A number of methods on Gen won't line up with old NICTA/rng names so there'll be some breakage there.)

You now have more control over the size of subsets of your random data.

Unspecified: (eg. .list, .string, etc)

scala> Gen.chooseInt(10).list.samples().take(5).foreach(println)
List(2, 7, 2, 6, 0, 0, 1, 3, 2, 2, 7, 0, 3, 3, 1, 1, 3, 2, 4, 8, 3, 0, 4, 6, 5, 3, 1)
List(2, 4, 3, 9, 9, 1, 4, 9, 0, 6, 2, 3, 2, 6, 8, 9, 2, 3, 8)
List(0, 5, 2, 7, 7, 3, 8, 6, 1, 5, 0, 3, 9, 2, 9, 8, 2, 5, 0, 2, 5, 8)
List(4)
List(0, 5, 0, 2, 2, 0, 0, 9, 1, 4, 4, 8, 4, 6, 9, 0)

Exact: (eg. .list(3), .string(10), etc)

scala> Gen.chooseInt(10).list(3).samples().take(5).foreach(println)
List(2, 2, 6)
List(5, 3, 1)
List(5, 5, 6)
List(1, 0, 4)
List(3, 9, 3)

Range: (eg. .list(0 to 3), .string(4 to 8), etc)

scala> Gen.chooseInt(10).list(0 to 3).samples().take(5).foreach(println)
List(4)
List()
List(3, 9, 7)
List(1, 4)
List(6)

It is also easier to use. Build up your generator, call .samples() and you have an infinite stream of data. Example:

scala> import nyaya.gen._
import nyaya.gen._

scala> val gen = Gen.int mapTo Gen.boolean.option
gen: nyaya.gen.Gen[Map[Int,Option[Boolean]]] = Gen(<function1>)

scala> gen.samples().take(1).foreach(println)
Map(609117252 -> None, --1220912339 -> Some(true), 1684851879 -> Some(false), 783799927 -> None)

Modularisation.

The previous two modules core and test have now been expanded into:

Module Package Contents
nyaya-prop nyaya.prop Proposition expression and evaluation.
nyaya-gen nyaya.gen Random data generation.
nyaya-test nyaya.test Prop testing and proving.
nyaya-util nyaya.util A few utilities used in the modules above.

Additionally the root package has changed from japgolly.nyaya to just nyaya.

Automate the transition with:

find . -type f -name '*.scala' -exec perl -pi -e 's/(?<![\w.])japgolly\.(?=nyaya)//g' {} +

*[[commit log]](https://github.com/japgolly/nyaya/compare/v0.5.12...v0.6.0)*