> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of Chris Toshok
>
> Paul Fisher wrote:
> >
> > Chris Toshok <[EMAIL PROTECTED]> writes:
> >
> > > At some level, purity of implementation has to give way to the
> > > utility of the implementation. I mean, it'll be nice when classpath
> > > works, but what happens if lots of code out there won't work with
> > > it?
> >
> > Purity does not need to give way to utility. GNU Classpath will run
> > programs which conform to Sun's public API. If someone has written
> > code which depends on undocumented classes/methods/fields, then that's
> > their fault.
>
> Are you sure you want to take this attitude? It seems a bit heavy
> handed. I mean, you're providing a service to developers. Why not make
> it easier for them to do their work?
>
> > Once the ISO spec for Java rolls around, I seriously doubt that the
> > only two class libraries in existence will be ours and Sun's.
>
> And they will all have to deal with this. Remember, Motif was an X/Open
> spec, and we still had to support the use of their protected fields.
>
If (and I don't know that this is true) Sun said *only public classes* are
supported on all Java implementations, then that would be sufficient for a
Classpath implementation. One of Java's primary purposes is portability, so
most programs are likely not to rely on undocumented features. Java was
*designed* so that you don't have to do that. You can't even get your
program certified 100% Java unless it uses only documented APIs.
If it isn't in the docs, people shouldn't be using it. Sun has taken pains
to ensure that people don't use (and don't have to use) undocumented and
unsupported features.
--John Keiser
