Hi Stephen, Thanks for testing this. Turns out I had removed a #:opaque in Definition while cleaning up the example before sending the mail. Now I get the same error as you.
With (define-syntax-class Definition #:opaque #:literals (define) (pattern (define (name:Id a:Id ...) body:Body))) I get the wrong error (because #:opaque "shortens the progress"). I had experimented with ~! as in: (define-syntax-class Definition #:opaque #:literals (define) (pattern (define ~! (name:Id a:Id ...) body:Body))) but the #:opaque ruins the effect of ~! here (not sure why). Sorry for the noise, Jens Axel 2015-07-25 16:55 GMT+02:00 Stephen Chang <stch...@ccs.neu.edu>: > Actually, I dont get the "begin expected" error that you mention. I'm > getting: > > define: expected identifier > parsing context: > while parsing Id > while parsing Definition > while parsing TopLevelForm in: 4 > > with the 4 highlighted, which seems correct since it comes from using > the Definition class? > > On Sat, Jul 25, 2015 at 10:51 AM, Stephen Chang <stch...@ccs.neu.edu> > wrote: > > I haven't played with the example in detail, but do you want the cut > > operator here? > > > > > http://docs.racket-lang.org/syntax/stxparse-patterns.html?q=syntax-parse#%28form._%28%28lib._syntax%2Fparse..rkt%29._~7e!%29%29 > > > > On Sat, Jul 25, 2015 at 9:55 AM, Jens Axel Søgaard > > <jensa...@soegaard.net> wrote: > >> Hi All, > >> > >> The syntax-parse form can often generate sensible errors based on the > >> progress made > >> during the attempt to match a pattern against a syntax-object. Given > >> multiple choices > >> the choice where the "maximum" progress was made is used to generate the > >> error message. > >> > >> Is there a way to tell syntax-parse, that a given choice is to be used > as > >> the source > >> of the error, if the match fails? > >> > >> A concrete example: In the toy language below a top-level form can be a > >> definition > >> or an expression. The malformed definition (define (foo 3) x) is not a > legal > >> definition -- it is however clear that the user intended to write a > >> definition, > >> since the form begins with (define ...). During matching syntax-parse > >> makes more progress under the assumption that the definition is a > sequence > >> expression, so the error message becomes: begin expected. > >> > >> How can I stop the matching prematurely when the (define ...) is seen? > >> > >> /Jens Axel > >> > >> #lang racket > >> ;;; > >> ;;; URLANG > >> ;;; > >> > >> ; In the grammar below x stands for a non-keyword identifier > >> > >> ; <program> ::= <top-level-form> ... > >> ; <top-level-form> ::= <definition> | <expr> > >> > >> ; <definition> ::= (define (x x ...) <body>) > >> ; <body> ::= <expr> > >> > >> ; <expr> ::= <datum> | <reference> | <application> | > <sequence> > >> ; <reference> ::= x > >> ; <application> ::= (x0 x ...) > >> ; <sequence> ::= (begin <expr> ...) > >> > >> ; <keyword> ::= define | begin > >> > >> ; <datum> ::= <fixnum> > >> > >> ; <identifier> an identifier that is not a keyword > >> ; <fixnum> an integer between -2^53 and 2^53 > >> > >> (require syntax/parse) > >> > >> (define min-fixnum (- (expt 2 53))) > >> (define max-fixnum (expt 2 53)) > >> > >> (define (Fixnum? r) > >> (and (number? r) (integer? r) > >> (<= min-fixnum r max-fixnum))) > >> > >> (define-syntax-class Fixnum > >> #:opaque > >> (pattern d > >> #:fail-unless (Fixnum? (syntax-e #'d)) #f)) > >> > >> (define-syntax-class Datum > >> (pattern (~or d:Fixnum))) > >> > >> (define-syntax-class Keyword > >> #:literals (begin define) > >> (pattern (~or begin define))) > >> > >> (define-syntax-class Id > >> (pattern (~and x:id > >> (~not y:Keyword)))) > >> > >> (define-syntax-class Reference > >> (pattern x:Id)) > >> > >> (define-syntax-class Application > >> #:literals (define) > >> (pattern (e:Expr ...))) > >> > >> (define-syntax-class Sequence > >> #:literals (begin) > >> (pattern (begin e0:Expr e:Expr ...))) > >> > >> (define-syntax-class Definition > >> #:literals (define) > >> (pattern (define (name:Id a:Id ...) body:Body))) > >> > >> > >> (define-syntax-class Body > >> (pattern b:Expr)) > >> > >> (define-syntax-class Expr > >> (pattern (~or e:Datum > >> e:Application > >> e:Reference > >> e:Sequence))) > >> > >> (define-syntax-class TopLevelForm > >> (pattern (~or t:Definition > >> t:Expr))) > >> > >> (define-syntax-class Program > >> (pattern (p:TopLevelForm ...))) > >> > >> ;;; The following expressions show some legal > >> ;;; constructs that are matched correctly. > >> > >> (syntax-parse #'3 > >> [d:Datum 'datum]) > >> > >> (syntax-parse #'3 > >> [e:Expr 'expr]) > >> > >> (syntax-parse #'(define (a x) x) > >> [t:TopLevelForm 'toplevelform]) > >> > >> (syntax-parse #'((define (foo x y) (+ x 1)) (foo 3)) > >> [p:Program 'program]) > >> > >> ;;; The malformed definition (define (a 4 x) x) is > >> ;;; correctly rejected as a toplevel form, but the > >> ;;; automatically generated error is not as intended. > >> > >> (syntax-parse #'(define (a 4 x) x) > >> [t:TopLevelForm 'toplevelform]) > >> > >> -- > >> You received this message because you are subscribed to the Google > Groups > >> "Racket Users" group. > >> To unsubscribe from this group and stop receiving emails from it, send > an > >> email to racket-users+unsubscr...@googlegroups.com. > >> For more options, visit https://groups.google.com/d/optout. > -- -- Jens Axel Søgaard -- You received this message because you are subscribed to the Google Groups "Racket Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to racket-users+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
