help-smalltalk
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Help-smalltalk] [RFC] Smalltalk scripting syntax


From: Paolo Bonzini
Subject: Re: [Help-smalltalk] [RFC] Smalltalk scripting syntax
Date: Wed, 14 Mar 2007 08:46:03 +0100
User-agent: Thunderbird 1.5.0.10 (Macintosh/20070221)

>> One would need something like
>>
>>         Object subclass: #MyClass!
>>         MyClass comment: 'A simple example class'!
>>
> Given the alternative I'd keep the exclamation point.  I can't imagine
> what would an interactive session look like.  I'm over my head here,
> though.  Come to think of it, Eval [] heavily reminds me of an ordinary
> function call.

The exclamation point is going to remain exactly for interactive
sessions.  Remember that -- anyway -- what's badly needed is, as Mike
pointed out earlier in this thread, nothing more than an easily parsable 
syntax that denotes the *structure* of the program rather than how to
reconstruct it.

That's the main reason why I am suspicious of syntaxes that are
pure Smalltalk.  It looks to me, though I might be wrong, that
having "Object subclass: #MyClass!" is as bad as having
"Object methodsFor: 'abc'!".  If somebody just adds a new method
like

    refine
        "Define a subclass of self, with the same name, in the current
         namespace."
        self environment = Namespace current ifTrue: [ self halt ].
        self subclass: self name!

how can you expect things to be parsed correctly when one writes:

    Set refine!
    Set comment: 'not your ordinary set'!

> But I'd
> rather choose that than using the temporary variable declaration
> literal.  I've gotten used to the idea that they're "temporary" and
> conceptually only exist during execution of something.  A lifetime of an
> object seems far too different to be called "execution" of the object.

Not if you consider a method's lifetime to be a context, and temporaries
to be instance variables of the context (they are in fact)...

And replying to Mike:

> I thought you might say that, but there's really no reason why you
> should insist on defining class instance variables in the class method
> scope.

I fail to parse this, sorry.  In fact, the bang syntax for class
instance variables is

    Foo class
        instanceVariableNames: 'uniqueInstance'!

which is "defined on the class" just like class methods.

In fact, the 1:1 mapping is *not* between instance and class variables,
but within instance and class-instance variables.

> The redundancy in the scope declaration and its message-send form.
> 
> Well, I can see that I'm going to have to agree to disagree.

Not at all.  For example, what if

    Class [
        ...
    ]

were a shortcut for

    Class ref: WhateverClassIsInScope class [
        ...
    ]

Then class methods and class instance variables would be just like

    Class [
        | uniqueInstance |
        uniqueInstance [ ^uniqueInstance ]
        uniqueInstance: anObject [ uniqueInstance := anObject ]
        new [
            self uniqueInstance ifFalse: [ self uniqueInstance: super new ]
            ^self uniqueInstance
        ]
    ]

Is the above proposal any better?

Paolo




reply via email to

[Prev in Thread] Current Thread [Next in Thread]