June 2007 - parenscript-devel - mailman3.common-lisp.net

[parenscript-devel] Miscellaneous utils I use
by John Fremlin 21 Jun '07

21 Jun '07

(defjsmacro max (a b) `(*Math.max ,a ,b)) ; might be nice to be able to have more than two arguments (defjsmacro ceil (expr) `(*Math.ceil ,expr)) (define-js-compiler-macro timeout (form timeout) (js-compile `(set-timeout (lambda() ,form) ,timeout))) (defjsmacro make-xml-http-request () `(if (slot-value window '*X-M-L-Http-Request ) (new *X-M-L-Http-Request) (new (*Active-X-Object "Microsoft.XMLHTTP"))))

2 5

[parenscript-devel] Feature request: speed (generating function)
by John Fremlin 19 Jun '07

19 Jun '07

In my dynamically generated webpages I regularly compile lots of very simple parenscript like `(append-markup ,element-id ,markup) where element-id and markup are Lisp strings. Parenscript takes a lot of time to do this in comparison to everything else. Ideally, I guess would be something that would generate a function like (defun js-append-markup (element-id markup) (output "appendMarkup(\"") (output (escape element-id)) (output "\",\"") (output (escape markup)) (output "\");"))

1 0

[parenscript-devel] Re:Js macro names can't be used as variable names
by Red Daly 17 Jun '07

17 Jun '07

I am away from a lispy computer, but here are the problem areas in js.lisp: (defun js-expand-form (expr) "Expand a javascript form." (cond ((atom expr) (multiple-value-bind (js-macro macro-env) (lookup-macro expr) (if js-macro (js-expand-form (let ((*js-macro-env* macro-env)) (funcall js-macro))) expr))) ;;; macros (define-js-compiler-macro macrolet (macros &rest body) (let* ((macro-env (make-hash-table :test 'equal)) (*js-macro-env* (cons macro-env *js-macro-env*))) (dolist (macro macros) (destructuring-bind (name arglist &rest body) macro (setf (gethash (symbol-name name) macro-env) (compile nil `(lambda ,arglist ,@body))))) (js-compile `(progn ,@body)))) (defjsmacro symbol-macrolet (macros &rest body) `(macrolet ,(mapcar #'(lambda (macro) `(,(first macro) () ,@(rest macro))) macros) ,@body)) The problem is that symbol-macrolets use the same machinery as regular macrolets. A solution is to maintain parallel symbol and regular macro environments... ;; new variables (eval-when (:compile-toplevel :load-toplevel :execute) (defvar *js-symbol-macro-toplevel* (make-hash-table :test 'equal) "Toplevel of symbol macro expansion, holds all the toplevel javascript macros.") (defvar *js-symbol-macro-env* (list js-symbol-macro-toplevel*) "Current symbol macro environment.")) ;; amend (lookup-macro ..) to take a :type argument. (defun lookup-macro (name &keyword (type :lambda)) "Lookup the macro NAME in the current macro expansion environment. Returns the macro and the parent macro environment of this macro." (unless (symbolp name) (return-from lookup-macro nil)) (do ((env (case type (:lambda *js-macro-env*) (:symbol *js-symbol-macro-env*) (t (error "Invalid macro type ~A" type))) (cdr env))) ((null env) nil) (let ((val (gethash (symbol-name name) (car env)))) (when val (return-from lookup-macro (values val (or (cdr env) (list (case type (:lambda *js-macro-toplevel*) (:symbol *js-symbol-macro-toplevel*) (t (error "Invalid macro type ~A" type)))))))))) ;; get rid of (defjsmacro symbol-macrolet ...) and replace it with ;; a compiler macro: (define-js-compiler-macro symbol-macrolet (macros &rest body) (let* ((macro-env (make-hash-table :test 'equal)) (*js-symbol-macro-env* (cons macro-env *js-symbol-macro-env*))) (dolist (macro macros) (setf (gethash (symbol-name (first macro)) macro-env) (compile nil `(lambda () ,@(rest macro))))) (js-compile `(progn ,@body)))) ;; change expand-form (defun js-expand-form (expr) "Expand a javascript form." (cond ((atom expr) (multiple-value-bind (js-macro macro-env) (lookup-macro expr :type :symbol) (if js-macro (js-expand-form (let ((*js-symbol-macro-env* macro-env)) (funcall js-macro))) expr))) ((js-compiler-macro-form-p expr) expr) ((equal (first expr) 'quote) expr) (t (let ((js-macro (lookup-macro (car expr)))) (if js-macro (js-expand-form (apply js-macro (cdr expr))) expr))))) This should at least give an idea of the relevant areas of code. -Red

1 0

[parenscript-devel] Re: [Bese-devel] ParenScript project page on common-lisp.net
by Luca Capello 16 Jun '07

16 Jun '07

Hello! On Fri, 15 Jun 2007 20:42:44 +0100, Vladimir Sedach wrote: > After a few delays, the new project page for ParenScript is up on > common-lisp.net: > > http://common-lisp.net/project/parenscript/ Sorry for the lag, I was busy driving to Edinburgh [1]. I'm (finally) planning a new Debian package for ParenScript and I'll more closely follow its upstream development. Thx, bye, Gismo / Luca Footnotes: [1] where the DebConf7 is ongoing: http://debconf7.debconf.org

1 0

[parenscript-devel] Feature request: Minify
by Daniel Gackle 16 Jun '07

16 Jun '07

I'd like to be able to set a special variable and have Parenscript produce minified output. That is, the resulting javascript would have no superfluous whitespace, the shortest possible symbol names, and so on. Don't care about obfuscation, though. Daniel

1 0

[parenscript-devel] Feature proposal: Optimize away constant arithmetic
by Vladimir Sedach 16 Jun '07

16 Jun '07

This is a feature I'm planning to add to ParenScript that involves eager evaluation for arithmetic constants, so for example: (js (+ foo.bar 0)) => "foo.bar;" (js (+ foo.bar 1 -1)) => "foo.bar;" (js (+ foo.bar 2 3)) => "foo.bar + 5;" This would make the generated Javascript be a few bytes shorter, but more importantly it seems that many Javascript implementations don't do constant arithmetic optimizations themselves, so it would speed up the generated Javascript as well.

1 0