Yes, outside in. Inside out doesn't work because the expander only knows where the inside is when it sees a core (fully expanded) form.
Robby On Wed, Dec 8, 2010 at 1:44 AM, YC <yinso.c...@gmail.com> wrote: > Hi all - > I have now implemented the pattern matching/transforming algorithm of a > macro expander (no hygiene yet) for a single macro expansion call. The next > step is to figure out how to apply the macro expanders recursively "in > order". > For example, let's say we want to expand the following: > > (or (and #t #t) #f) > > Should it be expanded via the head position first (i.e. finish expanding or > before expanding and) > > (or (and #t #t) #f) > => (let ((g (and #t #t))) (if g g #f)) > => ((lambda (g) (if g g #f)) (and #t #t)) > => ((lambda (g) (if g g #f)) (let ((h #t)) (if h #t #f))) > => ((lambda (g) (if g g #f)) ((lambda (h) (if h #t #t)) #t)) > > Or should the arguments be expanded before the head position expanding > again? (expanding or and and together): > > (or (and #t #t) #f) > => (let ((g (and #t #t))) (if g g #f)) > => (let ((g (let ((h #t)) (if h #t #f)))) (if g g #f)) > => ((lambda (g) (if g g #f)) (let ((h #t)) (if h #t #f))) > => ((lambda (g) (if g g #f)) ((lambda (h) (if h #t #f)) #t)) > > Does (and should) it matter either way? Any inputs are appreciated. > Thanks, > yc > On Sun, Dec 5, 2010 at 11:21 AM, YC <yinso.c...@gmail.com> wrote: >> >> >> On Sun, Dec 5, 2010 at 10:47 AM, Jon Rafkind <rafk...@cs.utah.edu> wrote: >>> >>> Attached is a simple macro expander (but not hygienic) that follows the >>> algorithm laid out in the paper "Macro by Example" by Wand and others. Its >>> written in python but the code follows the paper reasonably close so maybe >>> you can understand it. The paper is on the acm digital library if you can >>> load it (but for some reason its not loading for me right now, so here is a >>> google search link to it). >>> >>> >>> http://www.google.com/url?sa=t&source=web&cd=1&ved=0CBgQFjAA&url=http%3A%2F%2Fportal.acm.org%2Fcitation.cfm%3Fid%3D41632&ei=u9z7TPWCMY76sAO__MD3DQ&usg=AFQjCNGnasAUQsGpiVJik29wZRUwHBnz1w >>> >> >> Thanks Jon for the example code and the links to paper. I don't have >> access to ACM, but the python code looks quite understandable, even though I >> am not too familiar with Python. I will study the code to see what I can >> glean from it. >> > > _________________________________________________ > For list-related administrative tasks: > http://lists.racket-lang.org/listinfo/users > _________________________________________________ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/users
