[squeak-dev] The Inbox: Kernel-nice.1120.mcz

Nicolas Cellier nicolas.cellier.aka.nice at gmail.com
Mon Nov 13 21:30:14 UTC 2017


2017-11-13 22:29 GMT+01:00 Nicolas Cellier <
nicolas.cellier.aka.nice at gmail.com>:

>
>
> 2017-11-13 22:25 GMT+01:00 <commits at source.squeak.org>:
>
>> Nicolas Cellier uploaded a new version of Kernel to project The Inbox:
>> http://source.squeak.org/inbox/Kernel-nice.1120.mcz
>>
>> ==================== Summary ====================
>>
>> Name: Kernel-nice.1120
>> Author: nice
>> Time: 13 November 2017, 10:24:19.857289 pm
>> UUID: 24277641-be23-40ea-85ef-1db0d48f63d3
>> Ancestors: Kernel-mt.1119
>>
>> 1) Use // in Fraction>>gcd:, rather than / will was invoking the same
>> gcd: computation 4 times!
>>
> argh, "which was invoking", I did not finish to rephrase :(
>
> 2) Enhance the Fraction comment
>>
> This is in inbox because a class comment is something difficult to make
> right and will benefit peer advices.
>
benefit from, but it seems my hands are a bit decoupled from my brain this
evening...

> Feel free to improve!
>
>
>> The Fraction comment SHALL tell about the expected class invariants.
>> At least, it should help answering questions like:
>>
>> https://stackoverflow.com/questions/46942103/squeak-smalltal
>> k-why-sometimes-the-reduced-method-doesnt-work
>>
>> https://stackoverflow.com/questions/46905203/squeak-smalltal
>> k-why-reduction-of-a-fraction-does-not-happen-after-numerator-an
>>
>> While at it, also tell why 3 isFraction answers true, and 3.0 asFraction
>> -> an Integer, not a Fraction.
>> VW (st80?) has chosen better #isRational and #asRational messages for
>> making things a bit more clear, but without a Rational superclass, it's not
>> that obvious...
>>
>> =============== Diff against Kernel-mt.1119 ===============
>>
>> Item was changed:
>>   Number subclass: #Fraction
>>         instanceVariableNames: 'numerator denominator'
>>         classVariableNames: ''
>>         poolDictionaries: ''
>>         category: 'Kernel-Numbers'!
>>
>> + !Fraction commentStamp: 'nice 11/13/2017 22:09' prior: 0!
>> + Fraction provides methods for dealing with fractions like 1/3 as a
>> ratio of two integers (as apposed to a decimal representation 0.33333...).
>> - !Fraction commentStamp: '<historical>' prior: 0!
>> - Fraction provides methods for dealing with fractions like 1/3 as
>> fractions (not as 0.33333...).  All public arithmetic operations answer
>> reduced fractions (see examples).
>>
>> + instance variables:
>> +       numerator       <Integer> the number appearing before the
>> fraction bar (above)
>> +       denominator     <Integer> the number appearing after the fraction
>> bar (below)
>> +
>> + A Fraction is generally created by sending the message / to an Integer,
>> like in
>> - instance variables: 'numerator denominator '
>>
>> +     1 / 3
>> - Examples: (note the parentheses required to get the right answers in
>> Smalltalk and Squeak):
>>
>> + Alternatively, it is possible to create a new instance of Fraction by
>> sending #numerator:denominator: to the class.
>> + In this later case, it is then user responsibility to ensure that it
>> conforms to the following invariants:
>> +
>> + - the denominator shall allways be positive.
>> +   A negative Fraction shall have a negative numerator, never a negative
>> denominator.
>> +   Example: 1 / -3 will return -1/3
>> + - the denominator shall allways be greater than 1.
>> +   A Fraction with denominator 1 shall be reduced to its numerator (an
>> Integer).
>> +   Example 3 / 1 will answer 3 (the Integer) not 3/1
>> + - the numerator and denominator shall never have common multiples.
>> +   Common multiples shall allways be simplified until (numerator gcd:
>> denominator) = 1.
>> +   Example 8 / 6 will answer 4 / 3, because both 8=2*4 and 6=2*3 are
>> both divisible by 2.
>> +
>> + A Fraction non conforming to above invariants could be the cause of
>> undefined behavior and unexpected results.
>> + If unsure, it is advised to send the message #reduced to the freshly
>> created instance so as to obtain a conforming Fraction, or an Integer.
>> +
>> + Note that Fraction and Integer represent together the set of Rational
>> numbers:
>> + - Integer is a subset of rational (those which are whole numbers)
>> + - Fraction is used for representing the complementary subset of
>> rational (those which are not whole numbers)
>> +
>> + There could have been a Rational superclass to both Integer and
>> Fraction, and a message #isRational for testing if a Number is a Rational,
>> as well as a message #asRational for converting a Number to a Rational.
>> + But this level of indirection is not strictly necessary: instead, the
>> notion of Rational and Fraction are collapsed in Squeak, and Integer are
>> considered as a sort of special Fraction with unitary denominator.
>> + Thus #isFraction is the testing message, to which every Integer will
>> answer true, since considered as a sort of Fraction.
>> + And #asFraction is the conversion message, that may answer an instance
>> of Fraction of Integer, depending if the corresponding rational number is
>> whole or not.
>> +
>> + All public arithmetic operations will answer reduced fractions.
>> + Examples:
>> +
>>   (2/3) + (2/3)
>> + (2/3) + (1/2)         "case showing reduction to common denominator"
>> + (2/3) + (4/3)         "case where result is reduced to an Integer"
>> + (2/3) raisedToInteger: 5               "fractions also can be
>> exponentiated"
>> - (2/3) + (1/2)          "answers shows the reduced fraction"
>> - (2/3) raisedToInteger: 5               "fractions also can have
>> exponents"
>>   !
>>
>> Item was changed:
>>   ----- Method: Fraction>>gcd: (in category 'arithmetic') -----
>>   gcd: aFraction
>>         | d |
>>         d := denominator gcd: aFraction denominator.
>> +       ^(numerator *(aFraction denominator//d) gcd: aFraction
>> numerator*(denominator//d)) / (denominator//d*aFraction denominator)!
>> -       ^(numerator *(aFraction denominator/d) gcd: aFraction
>> numerator*(denominator/d)) / (denominator/d*aFraction denominator)!
>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20171113/1ceb0947/attachment.html>


More information about the Squeak-dev mailing list