About KCP and automatic initialize
ducasse
ducasse at iam.unibe.ch
Wed Sep 10 07:05:32 UTC 2003
Just for your information in my new book I refused to go in that trap
so I modified the system
and just say the truth at chapter 55.
BTW ANSI is crapy bad it is not a standard, it is the result of vendors
pushing all in their directions.
testing the presence of protocol is idiot. So what do we want to keep
this standard that cover nothing expect hilarious points, seriously I
dream about R5RS. This is a standard a real one.
Stef
On Mercredi, sep 10, 2003, at 06:14 Europe/Zurich, Richard A. O'Keefe
wrote:
> This topic has been discussed before.
>
> I have a lot of sympathy for it, BUT
> one of the standard beginner (and not-so-beginner) complaints about
> Squeak is the documentation. (Yes, there is a lot of documentation.
> It's all over the place, it's often hard to know where to look, and
> up-to-date is not a phrase one could honestly use to describe it all.)
> Where do you turn? Well, there are all those free Smalltalk books.
>
> What happens if you break with the past so violently in something
> as basic as how you create an object? What use are the Smalltalk books
> going to be to beginners who can't even trust them to be applicable to
> Squeak in such a basic way?
>
> The ANSI Smalltalk standard says nothing about any #initialize method
> being called by #new; there is no #initialize in Object.
>
> If you just want to make life easier for people constructing new
> classes,
>
> Object subclass: #AutoInitObject
> methods
> initialize
> ^self
> class methods
> new
> ^super new initialize
>
> and then people can make subclasses of AutoInitObject instead of
> Object.
> So we don't _need_ any change to the kernel classes to get most of the
> benefit, and we'd lose far too much in the way of applicable
> documentation
> (books, tutorials, examples) and existing code if such a change were
> made
> to Object.
>
>
More information about the Squeak-dev
mailing list
|