PLT Scheme 4.0 to introduce major new changes to Scheme | Hacker News Reader

PLT Scheme 4.0 to introduce major new changes to Scheme

blog.plt-scheme.org

45 pointslurkage15 comments

Unfortunate title. PLT Scheme 4.0 introduces major new changes to PLT Scheme.

"Improvements to the PLT Scheme language include better syntax for modules, better support for optional and keyword function arguments, more expressive syntax for structure types, streamlined hash-table operations, new syntax for list comprehensions and iterations, a more complete and consistent set of list and string operations, and reduced dependence on mutable pairs."

In the abstract, sounds like a Scheme that can compete with Python and Ruby.

One can only hope. With all the furor over the quality of interpreters that SquirrelFish has kicked off in the community, the MzScheme interpreter is looking pretty good. It's extremely fast and also very lean, and the plt-scheme is a very good language with a competent FFI, multi-platform support, and a lot of advanced features.

"New syntax" doesn't necessarily mean replacing parens with something else. I wouldn't count on a popularity boost from any of this. (I assume by "compete" you mean in terms of popularity, since that's the only area where all Scheme implementations are behind Python and Ruby.)

I've been using the pre-release versions for a while now and there are some nice improvements. Some of the syntactic sugar is really nice to use, like the new module and require forms, and the generalized lambda forms (combined lambda, opt-lambda, define/kw,...).

Does anyone else think that PLT Scheme dialect is moving closer to Common Lisp?

Stone soup.

Quick summary:

A stranger walks into a village, puts a pot on a fire, fills it with water and, in full view of the villagers, drops in a single stone. Curious, the villagers ask what he's up to.

"Making stone soup."

"You can make soup from a stone?"

"Yes, it's very good. However, it's even better with a little bit of..."

You see how this is going to go. One by one, the villagers bring some little thing or other to add to the stone soup, and in the end they all enjoy some, shocked at how good a soup made from a stone can taste.

The relevance: The whole point of Scheme is to be a minimalist programming language. However, then someone says, "Yes, I love this minimal programming language. Very elegant. However, if you just had support for hash tables..."

After several iterations of this, you have users of this programming language saying "Amazing how much you can do with this minimalist programming language..."

Good regurgitation of Agile/Pragmatic/... folklore. Pity it has no base in reality.

Most Scheme implementations have always had functionality well beyond the standard, and many Scheme users don't see a contradiction in having a minimalistic language standard[1] and rich implementations.

[1] I'm talking about R5RS here, but the story doesn't apply to R6RS either. That was never intended to be a small incremental step, as your fable suggests.

P.S.: BTW, I think the metaphor you meant was the boiled frog, rather than stone soup.

No, it doesn't happen that way.

It's SICP that amazes people at what can be done with a minimalist programming language.

People who use a Scheme implementation with lots of additions have no illusions of minimalism.

im watching the videos from/about sicp now:

http://www.swiss.ai.mit.edu/classes/6.001/abelson-sussman-le...

amazing stuff, recommend everyone to watch.

I was under the impression scheme is still minimalist and that plt/scheme is as your describe. The story is nice, but I think it's incorrectly applied.

PLT Scheme still doesn't have a meta-object protocol, right? So it's still a ways behind Common Lisp the language, in my opinion. However, as far as the implementation goes, PLT Scheme is really nice and fully-featured. It also has one feature I like a whole lot: it's fully embeddable. As far as I know, neither SBCL, nor Allegro CL, nor LispWorks, nor Clozure are (I know ECL is, but I also don't know anyone who uses it). This makes PLT Scheme a reasonable choice for something like game scripting.

Sure, and that's okay. A lot of people like the Lisp-1 rules and PLT's interpreter is technically very good.

Scheme as a spec is a simple, minimalist language. Scheme as an implementation need not be this. Very few people can fit their problem into pure, unaugmented r5rs scheme. Even if they can, it may mean writing a lot more code than other languages.

I, for one, am excited to see the scheme community embrace a pragmatic goal as opposed to a purist one. Just because they support some reasonable things like FFI and keyword arguments from the get-go doesn't mean that they're suddenly common lisp with a loop sublanguage and iteration in their printfs.

I just did a little wandering around the available documentation on PLT-Scheme and noticed that large pieces are untouched since 2004. Other large pieces seem to be missing. Seems like an un-propitious start for 4.0! Or perhaps this is what they meant by new work for 4.0 docs...

try here: http://pre.plt-scheme.org/docs/html/

Reply on news.ycombinator.com