The Trunk: SUnit-eem.91.mcz

> Eliot Miranda uploaded a new version of SUnit to project The Trunk:
> http://source.squeak.org/trunk/SUnit-eem.91.mcz
>
> ==================== Summary ====================
>
> Name: SUnit-eem.91
> Author: eem
> Time: 5 February 2013, 9:45:22.157 pm
> UUID: 563010a6-5bbc-4312-ae7c-b09410b24fad
> Ancestors: SUnit-cmm.90
>
> Robust testCase printing.  i.e. TestCase new asString should
> not raise an error.
>
> =============== Diff against SUnit-cmm.90 ===============
>
> Item was changed:
>  ----- Method: TestCase>>printOn: (in category 'printing') -----
>  printOn: aStream
>
> + testSelector
> + ifNil: [super printOn: aStream]
> + ifNotNil:
> + [aStream
> + nextPutAll: self class printString;
> + nextPutAll: '>>#';
> + nextPutAll: testSelector] !
> - aStream
> - nextPutAll: self class printString;
> - nextPutAll: '>>#';
> - nextPutAll: testSelector
> - !
>
>
>

> Kent Beck makes a case for using #print: in his Reversing Method
> pattern on page 33 of the original "Smalltalk Best Practice Patterns",
> but the basis for that is to avoid having messages going to different
> receivers on every line.
>
> If Eliot had used
>
>  self class printOn: aStream.
>  aStream nextPutAll: '>>#'.
>  testSelector printOn: aStream
>
> then Kents argument would apply.  However, since Eliot has a nice
> cascaded flow of messages all going to the same receiver, the
> Reversing Method isn't necessary.
>
> Having said that, I normally prefer not to send #printString to any
> object inside a printOn: method -- since that object should just be
> able to print itself on the stream.  So I think print: would be nicer:
>
>  aStream
>     print: self class;
>     print: '>>#';
>     print: testSelector
>
> BUT, having said THAT, the caveat with #print: comes with Strings and
> Characters -- that they print the syntax punctuation; strings
> surrounded by single-quotes and Characters prefixed with '$' rather
> than simply printing the contents like nextPutAll:.