Tolerance matchers fail on edge-case numeric inputs

[erik]

While working on #3096, the property-based checker found some edge cases for tolerance matchers which don't appear to work correctly (or at least have somewhat strange behavior).

Cases where we should probably throw an IllegalArgumentException:

(Float.NaN).shouldBeWithinPercentageOf(Float.NaN, 1.0)         // => NaN should be in NaN..NaN
(Double.NaN).shouldBeWithinPercentageOf(Double.NaN, 1.0)       // => NaN should be in NaN..NaN
0.shouldBeWithinPercentageOf(0, Double.POSITIVE_INFINITY)      // => 0 should be in NaN..NaN
0L.shouldBeWithinPercentageOf(0L, Double.POSITIVE_INFINITY)    // => 0 should be in NaN..NaN
0f.shouldBeWithinPercentageOf(0f, Double.POSITIVE_INFINITY)    // => 0.0 should be in NaN..NaN
0.0.shouldBeWithinPercentageOf(0.0, Double.POSITIVE_INFINITY)  // => 0.0 should be in NaN..NaN

Cases where behavior differs from x shouldBe (y plusOrMinus z.percent):

// Applies to both Double and Float

NEGATIVE_INFINITY shouldBe (NEGATIVE_INFINITY plusOrMinus 1.percent) // OK
POSITIVE_INFINITY shouldBe (POSITIVE_INFINITY plusOrMinus 1.percent) // OK

NEGATIVE_INFINITY.shouldBeWithinPercentageOf(NEGATIVE_INFINITY, 1.0) // -Infinity should be in -Infinity..NaN
POSITIVE_INFINITY.shouldBeWithinPercentageOf(POSITIVE_INFINITY, 1.0) // Infinity should be in NaN..Infinity

Which version of Kotest are you using #3096 - branched from 56573a7

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[LeoColman]

Nice catch!

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[sksamuel]

Was fixed in 5.6.0