Name: Anonymous 2011-01-26 13:46
Well, it's about the same as Gambit-C or Chicken in speed, both interpretation & compilation, but I've added other things to it:
- dict's (backed by tries) with {} literal syntax.
- dicts have a partial-key? semi-predicate which returns #t on match, #f on no match and length of match if it is a partial match. I've used a similar method for search systems before, so I thought it would be useful to have in-language.
- vectors with [] literal syntax.
- unified collection API (cset!,cupdate,length,nth,cslice,&c.).
- collections are functions whose domain is their keys and codomain is the storage field
- case-lambdas are there by default.
- support for "strict enough" typing, and type inference (which I'm still working on).
- extensions to syntax-rules to match against type.
- extensions to various SRFIs.
- Heavy POSIX integration that is abstracted away into modules.
- large library (image processing, network protocols, file formats, &c. still more to be added) out of the box.
and various others. I want it to be ready "out of the box" for most work, like Python, but still small enough that you can carry it around with you from system to system (the interpreter is something like 11k lines of code, which is the run time, the core interpreter, the FFI, the POSIX integration & some cruft; I'm hoisting the system so that it can be fully self-hosting, but that's a work in progress; for instance, I'm still working on the GC). The binary for the interpreter is about 166k in size. So, I wanted to "do it right" for most of the uses that I have for Scheme now.
edit: I also forgot to mention that it has W7 in the core (implemented as primitives) for security, plus Clojure-style metadata on all SExpressions (and a W7 environment to include taint checks).
Found here: http://www.reddit.com/r/scheme/comments/f73bu/what_is_in_your_opinion_the_best_scheme/c1dsz8v
Thoughts?
- dict's (backed by tries) with {} literal syntax.
- dicts have a partial-key? semi-predicate which returns #t on match, #f on no match and length of match if it is a partial match. I've used a similar method for search systems before, so I thought it would be useful to have in-language.
- vectors with [] literal syntax.
- unified collection API (cset!,cupdate,length,nth,cslice,&c.).
- collections are functions whose domain is their keys and codomain is the storage field
- case-lambdas are there by default.
- support for "strict enough" typing, and type inference (which I'm still working on).
- extensions to syntax-rules to match against type.
- extensions to various SRFIs.
- Heavy POSIX integration that is abstracted away into modules.
- large library (image processing, network protocols, file formats, &c. still more to be added) out of the box.
and various others. I want it to be ready "out of the box" for most work, like Python, but still small enough that you can carry it around with you from system to system (the interpreter is something like 11k lines of code, which is the run time, the core interpreter, the FFI, the POSIX integration & some cruft; I'm hoisting the system so that it can be fully self-hosting, but that's a work in progress; for instance, I'm still working on the GC). The binary for the interpreter is about 166k in size. So, I wanted to "do it right" for most of the uses that I have for Scheme now.
edit: I also forgot to mention that it has W7 in the core (implemented as primitives) for security, plus Clojure-style metadata on all SExpressions (and a W7 environment to include taint checks).
Found here: http://www.reddit.com/r/scheme/comments/f73bu/what_is_in_your_opinion_the_best_scheme/c1dsz8v
Thoughts?