Sean E. Russell wrote:
> Yeah.  That is a different sort of backwards compatibility, though.  I'm 
> considering deprecating an entire *package*.  But, as I said, I'm still 
> trying to find a reasonable way through this.
>   

Is there some reason why a thin API facade is impossible? You wouldn't 
need to make it high performance.

> Java 1.0 -> 1.5 didn't break anything.  Java 1.0 code is legal 1.5 code.

Which is why Java has (for example) at least four different equivalents 
of a Ruby Array, all subtly incompatible with each other, and that's not 
counting stacks, sorted sets or lists. Some of us think that's a bad thing.

Java took the approach of not wanting anything to break ever, because 
that was a commercial selling point of the language--write your code 
once, and it'll run anywhere, implicit in which was the promise that it 
would keep running. Ruby being free open source software has the option 
of not chasing popularity, but instead chasing cleanliness.

Perl is another example of the "sounds good, throw it in" approach to 
language design. Python is the opposite extreme, "do it our way exactly, 
or get lost". Ruby seems to be somewhere in between, which probably 
means these conversations are going to be endless.


mathew