On Mon, Sep 10, 2012 at 6:46 AM, Robby Findler <ro...@eecs.northwestern.edu> wrote: > On Sun, Sep 9, 2012 at 6:52 PM, Matthew Flatt <mfl...@cs.utah.edu> wrote: >> Getting the line, column, and span for a rewriter result can be tricky. >> Usually, you want the result of a rewriter to have the same start and >> span as the input. I think I got that part right, below, too. > > with-compound-rewriter has some smarts about how to insert lw structs > when it sees non-lws in the resulting list. So you can exploit that by > writing something like this (and thus avoid the need for "collapse"): > > (define (dummy-rewriters thunk) > (with-compound-rewriters > (['n (lambda (lws) > (list "" > (list-ref lws 2) > (text (symbol->string (lw-e (list-ref lws 1))) > `(subscript . ,(default-style)) > (default-font-size)) > ""))]) > (thunk))) > > I've added a note about this trick to the docs. The behavior is not the same as Matthew Flatt's solution. With: (dummy-rewriters (lambda () (render-term dummy ( (n 112221111) (n 1223333))))) I get just one garbled line with this trick, while with Matthew Flatt's solution, I get two lines as expected. Is this bug in the trick's implementation?
In any case, a big thank you, since I've been able to get the ball rolling on my typesetting project. Best regards, ~n > > Robby ____________________ Racket Users list: http://lists.racket-lang.org/users
