Smalltalk › Squeak › Squeak - Dev

Character space and literal arrays

‹ Previous Topic Next Topic ›
 
classic Classic list List threaded Threaded
♦
♦
 3 messages Options
Eliot Miranda-2
Reply | Threaded
Open this post in threaded view
♦
♦
|

Character space and literal arrays

Eliot Miranda-2
Hi Lukas,

    there's a problem with the current printOn:/storeOn: distinction in current Squeak.  The issue is that within a literal Array all characters must be printed using the $ form, /not/ the Character message or (Character value: N) forms. e.g.

#($ ) storeString => #(Character space)

(Compiler evaluate: #($ ) storeString) size => 2

(Compiler evaluate: #($ ) storeString) = #(#Character #space) => true

In VisualWorks (7.7) the above works correctly (*)

#($ ) storeString => #($ )

(Compiler evaluate: #($ ) storeString) size => 1

(Compiler evaluate: #($ ) storeString) = #(#Character #space) => false

and works because Array asserts the "characters print literally" stream policy while printing literally.


I stumbled upon this looking at the decompiler tests, specifically of MethodPragmaTest>>testCompileCharacter:
testCompileCharacter
self assertPragma: 'foo: $a' givesKeyword: #foo: arguments: #( $a ).
self assertPragma: 'foo: $ ' givesKeyword: #foo: arguments: #( $  ).

which decompiled shows as
testCompileCharacter
self
assertPragma: 'foo: $a'
givesKeyword: #foo:
arguments: #($a ).
self
assertPragma: 'foo: $ '
givesKeyword: #foo:
arguments: #(Character space )



(*) but VW still has a bizarre def for printString, 

(Array with: Character space) printString '#(Core.Character space)'

(Compiler evaluate: (Array with: Character space) printString) size => 2


best
Eliot
Nicolas Cellier
Reply | Threaded
Open this post in threaded view
♦
♦
|

Re: Character space and literal arrays

Nicolas Cellier
something old like http://bugs.squeak.org/view.php?id=4322
http://bugs.squeak.org/view.php?id=5700

Nicolas

2010/9/22 Eliot Miranda <[hidden email]>:

> Hi Lukas,
>     there's a problem with the current printOn:/storeOn: distinction in
> current Squeak.  The issue is that within a literal Array all characters
> must be printed using the $ form, /not/ the Character message or (Character
> value: N) forms. e.g.
> #($ ) storeString => #(Character space)
> (Compiler evaluate: #($ ) storeString) size => 2
> (Compiler evaluate: #($ ) storeString) = #(#Character #space) => true
> In VisualWorks (7.7) the above works correctly (*)
> #($ ) storeString => #($ )
> (Compiler evaluate: #($ ) storeString) size => 1
> (Compiler evaluate: #($ ) storeString) = #(#Character #space) => false
> and works because Array asserts the "characters print literally" stream
> policy while printing literally.
>
> I stumbled upon this looking at the decompiler tests, specifically of
> MethodPragmaTest>>testCompileCharacter:
> testCompileCharacter
> self assertPragma: 'foo: $a' givesKeyword: #foo: arguments: #( $a ).
> self assertPragma: 'foo: $ ' givesKeyword: #foo: arguments: #( $  ).
> which decompiled shows as
> testCompileCharacter
> self
> assertPragma: 'foo: $a'
> givesKeyword: #foo:
> arguments: #($a ).
> self
> assertPragma: 'foo: $ '
> givesKeyword: #foo:
> arguments: #(Character space )
>
>
> (*) but VW still has a bizarre def for printString,
> (Array with: Character space) printString '#(Core.Character space)'
> (Compiler evaluate: (Array with: Character space) printString) size => 2
>
> best
> Eliot
>
>
>

Lukas Renggli
Reply | Threaded
Open this post in threaded view
♦
♦
|

Re: Character space and literal arrays

Lukas Renggli
In reply to this post by Eliot Miranda-2
Hi Eliot,

Indeed, this looks like a issue.

Interestingly the refactoring browser avoids the problem and
implements its own code for printing character literals.

So you suggest to replace Character>>#storeOn: with

Character>>#storeOn: aStream
   aStream nextPut: $$; nextPut: self

?

However this will break again if the character is not supported in aStream.

Lukas

On 22 September 2010 23:10, Eliot Miranda <[hidden email]> wrote:

> Hi Lukas,
>     there's a problem with the current printOn:/storeOn: distinction in
> current Squeak.  The issue is that within a literal Array all characters
> must be printed using the $ form, /not/ the Character message or (Character
> value: N) forms. e.g.
> #($ ) storeString => #(Character space)
> (Compiler evaluate: #($ ) storeString) size => 2
> (Compiler evaluate: #($ ) storeString) = #(#Character #space) => true
> In VisualWorks (7.7) the above works correctly (*)
> #($ ) storeString => #($ )
> (Compiler evaluate: #($ ) storeString) size => 1
> (Compiler evaluate: #($ ) storeString) = #(#Character #space) => false
> and works because Array asserts the "characters print literally" stream
> policy while printing literally.
>
> I stumbled upon this looking at the decompiler tests, specifically of
> MethodPragmaTest>>testCompileCharacter:
> testCompileCharacter
> self assertPragma: 'foo: $a' givesKeyword: #foo: arguments: #( $a ).
> self assertPragma: 'foo: $ ' givesKeyword: #foo: arguments: #( $  ).
> which decompiled shows as
> testCompileCharacter
> self
> assertPragma: 'foo: $a'
> givesKeyword: #foo:
> arguments: #($a ).
> self
> assertPragma: 'foo: $ '
> givesKeyword: #foo:
> arguments: #(Character space )
>
>
> (*) but VW still has a bizarre def for printString,
> (Array with: Character space) printString '#(Core.Character space)'
> (Compiler evaluate: (Array with: Character space) printString) size => 2
>
> best
> Eliot



--
Lukas Renggli
www.lukas-renggli.ch

Free forum by Nabble Edit this page