Why isn't Rev more popular?

[Mark Wieder]

Bill-

Friday, December 2, 2005, 12:01:56 PM, you wrote:

I'm with you most of the way on this. I'd love to have a way to
implement my own keywords into my stacks. And then distribute
libraries that extend xtalk with them.

Problem 1: most other languages that allow this are adopting namespace
conventions to deal with conflicting libraries. That's where you start
having to pick up the dot-notation or something similar, and veering
away from the near-natural-language approach of xtalk.

Problem 2: It's not just codesize bloat that you have to deal with
when extending the language, it's also slowing down the basic parsing
engine having to deal with more and more stuff. I'd really like to
avoid slowing down one of my stacks that doesn't use a function that
someone else wanted included.

Problem 3: <warning - heresy follows> there's no such thing as
"self-documenting, human-readable code". Sorry - that's the way it is.
Xtalk comes pretty close, though. Even comes pretty close to
"human-thinkable".

Problem 4: creeping featureitis

New things *do* get added to xtalk. But very slowly, and after much
deliberation. And I like it that way. It's hard to retract something
once you've added it to a language. I think the point at which you
move something from a library into the core functionality needs to be
looked at very cautiously, and runrev has, so far, exhibited a lot of
caution in this respect (to the point of ignoring several of my bug
report enhancement requests).

-- 
-Mark Wieder
 mwieder at ahsoftware.net