The Inbox: CollectionsTests-nice.338.mcz

Nicolas Cellier uploaded a new version of CollectionsTests to project The Inbox:
http://source.squeak.org/inbox/CollectionsTests-nice.338.mcz

==================== Summary ====================

Name: CollectionsTests-nice.338
Author: nice
Time: 2 May 2020, 12:46:51.581176 am
UUID: 9860b2e2-750f-da40-a73f-4332c3facd4f
Ancestors: CollectionsTests-mt.337

Introduce tests for Float64Array

=============== Diff against CollectionsTests-mt.337 ===============

Item was added:
+ ClassTestCase subclass: #Float64ArrayTest
+ instanceVariableNames: ''
+ classVariableNames: ''
+ poolDictionaries: ''
+ category: 'CollectionsTests-Arrayed'!
+
+ !Float64ArrayTest commentStamp: 'nice 5/30/2006 01:24' prior: 0!
+ These tests are used to assert that Float64ArrayPlugin has same results as Float asIEEE32BitWord!

Item was added:
+ ----- Method: Float64ArrayTest>>testArithmeticCoercion (in category 'tests') -----
+ testArithmeticCoercion
+
+ self should: [3.5 / (Float64Array with: 2.0) = (Float64Array with: 1.75)].
+ self should: [3.5 * (Float64Array with: 2.0) = (Float64Array with: 7.0)].
+ self should: [3.5 + (Float64Array with: 2.0) = (Float64Array with: 5.5)].
+ self should: [3.5 - (Float64Array with: 2.0) = (Float64Array with: 1.5)].!

Item was added:
+ ----- Method: Float64ArrayTest>>testFloat64ArrayPluginPrimitiveAt (in category 'tests') -----
+ testFloat64ArrayPluginPrimitiveAt
+ "Trivial test"
+
+ #(zero one epsilon fmax fminDenormalized fminNormalized negativeZero negativeInfinity infinity) do: [:e |
+ | aFloat |
+ aFloat := Float perform: e.
+ self assert: ((Float64Array with: aFloat) at: 1) = aFloat].
+
+ "Cannot compare NaN"
+ self assert: ((Float64Array with: Float nan) at: 1) isNaN.!

Item was added:
+ ----- Method: Float64ArrayTest>>testThatDivideAvoidsUnecessaryOverflow (in category 'tests') -----
+ testThatDivideAvoidsUnecessaryOverflow
+ | f1 f2 expectedResult |
+ f1 := Float fminNormalized.
+ f2 := Float fminDenormalized.
+ expectedResult := Float64Array with: f1 / f2.
+
+ "If any of these primitives inverse the dividend, then it will overflow"
+ self assert: expectedResult equals: f1 / (Float64Array with: f2).
+ self assert: expectedResult equals: (Float64Array with: f1) / f2.
+ self assert: expectedResult equals: (Float64Array with: f1) / (Float64Array with: f2).!

Item was added:
+ ----- Method: Float64ArrayTest>>testVectorOperations (in category 'tests') -----
+ testVectorOperations
+
+ "Test primtive cases 'receiver size = argument size'."
+ self assert: (Float64Array withAll: {2.0. 2.0}) equals: (Float64Array withAll: {4.0. 6.0}) / (Float64Array withAll: {2.0. 3.0}).
+ self assert: (Float64Array withAll: {8.0. 9.0}) equals: (Float64Array withAll: {4.0. 4.5}) * (Float64Array withAll: {2.0. 2.0}).
+ self assert: (Float64Array withAll: {6.0. 9.0}) equals: (Float64Array withAll: {4.0. 6.0}) + (Float64Array withAll: {2.0. 3.0}).
+ self assert: (Float64Array withAll: {2.0. 3.0}) equals: (Float64Array withAll: {4.0. 6.0}) - (Float64Array withAll: {2.0. 3.0}).
+ self assert: 26 equals: ((Float64Array withAll: {4.0. 6.0}) dot: (Float64Array withAll: {2.0. 3.0})).
+
+ "Test corner cases where 'receiver size < argument size'."
+ self should: [(Float64Array withAll: {4.0. 6.0}) / (Float64Array withAll: {2.0. 3.0. 42.0})] raise: Error.
+ self should: [(Float64Array withAll: {4.0. 6.0}) * (Float64Array withAll: {2.0. 3.0. 42.0})] raise: Error.
+ self should: [(Float64Array withAll: {4.0. 6.0}) + (Float64Array withAll: {2.0. 3.0. 42.0})] raise: Error.
+ self should: [(Float64Array withAll: {4.0. 6.0}) - (Float64Array withAll: {2.0. 3.0. 42.0})] raise: Error.
+ self should: [(Float64Array withAll: {4.0. 6.0}) dot: (Float64Array withAll: {2.0. 3.0. 42.0})] raise: Error.
+
+ "Test corner cases where 'receiver size > argument size'."
+ self should: [(Float64Array withAll: {4.0. 6.0. 42.0}) / (Float64Array withAll: {2.0. 3.0})] raise: Error.
+ self should: [(Float64Array withAll: {4.0. 6.0. 42.0}) * (Float64Array withAll: {2.0. 3.0})] raise: Error.
+ self should: [(Float64Array withAll: {4.0. 6.0. 42.0}) + (Float64Array withAll: {2.0. 3.0})] raise: Error.
+ self should: [(Float64Array withAll: {4.0. 6.0. 42.0}) - (Float64Array withAll: {2.0. 3.0})] raise: Error.
+ self should: [(Float64Array withAll: {4.0. 6.0. 42.0}) dot: (Float64Array withAll: {2.0. 3.0})] raise: Error.
+
+
+
+ !

Item was added:
+ ----- Method: FloatArrayTest>>testThatDivideAvoidsUnecessaryOverflow (in category 'tests') -----
+ testThatDivideAvoidsUnecessaryOverflow
+ | f1 expectedResult |
+ f1 := 1.0e-39.
+ expectedResult := Float64Array with: 1.0.
+
+ "If any of these primitives inverse the dividend into a float, then it will overflow.
+ It will work though if using intermediate double."
+ self assert: expectedResult equals: f1 / (Float64Array with: f1).
+ self assert: expectedResult equals: (Float64Array with: f1) / f1.
+ self assert: expectedResult equals: (Float64Array with: f1) / (Float64Array with: f1).!