> > The language is massively fragmented already, so this is nothing new. > Well, a few of the people who voted yes on the R6RS proposal seemed to believe that the new standard will bring about better compatibility between existing implementations. My point was that you cannot alter the existing state of Scheme by the sheer force of willpower.
> > I prefer the view of the ANSI/ISO C standards committee: "Old code > is important; old implementations are not." Even Sun has now been > dragged kicking and screaming into the ISO C age. I most certainly do not consider ISO C or C++ to be good language standards. In my opinion, a programming language standard should encourage and support the proliferation of high-quality implementations. For example, R5RS or the Standard ML specification. The follow very different philosophies: R5RS is short and laconic, and thus encourages experimentation; the Standard ML standard contains a full formal specification of the language semantics, and thus encourages consistency and correctness. Both languages enjoy a variety of feature-complete high-quality compilers. Compare with C++: despite the thousands of pages of specification, I am not convinced that you can find two different C++ compilers that implement all features of the language in the same way. Small specifications that only concern high-level semantics seem to encourage higher implementation quality. -Ivan _______________________________________________ Chicken-users mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/chicken-users
