Thanks. I was aware of the potential performance issues. However this is just an exercise so I'm not too concerned about scalability. We've been asking this as an interview question so I though it would be interesting to implement it in Pico Lisp.
Your suggested code includes a few functions I havn't looked at yet like 'make' and 'link', there not somthing I've come accross in other languages. Having two strings which look the same but are not equal seems like a bit of a design flaw. The intuitive notion would be that strings that look the same should always be equal. regs Konrad 2008/6/12 Alexander Burger <[EMAIL PROTECTED]>: > Hi Konrad, > >> Instead I get >> $385555801 NIL >> "orange" 1 >> "orange" 1 >> >> I don't understand why two invocations of (read T) are returning two > ^ > (a minor remark: You certainly meant 'line' here, instead of 'read') > > >> symbols which look the same but are not equal. Its somthing > > Yes, exactly. This effect results from the fact that the two symbols > subsequently read are not the same. And 'put', 'get' etc. search the > property list for pointer equality. > >> to do with dynamic scoping how do I get arround it. > > (line T) reads a line into a transient symbol. Transient symbols are > only the same if they are read within the same read-eval-loop (i.e. the > 'load' function). That is, if they are read within the scope of a single > source file. > > Transient symbols created by some other means (e.g. 'line', or 'pack' > and son on) are always unique. > > >> The goal is to count the number of files a particular word appears in. > > You could use association-lists instead of a symbol and 'put'. > Especially also because there is a function 'accu' which does the > counting: > > (de read-words (File-name) > (in File-name > (uniq > (make > (until (eof) > (link (line T)) > (uniq res) ) ) ) ) ) > > (de main () > (let Acum NIL > (for Fname (argv) > (for Word (read-words Fname) > (accu 'Acum Word 1) ) ) # 'accu' from "lib/misc.l" > Acum ) ) > > (another remark: I strongly recommend to use upper case names for > locally bound symbols, to avoid clashes with global symbols like > function names) > > > If you have a lot of words (say, more than 100), neither a property list > nor an association list is a good idea. I'd recommend a tree structure: > > (de main () > (let Acum NIL > (for Fname (argv) > (for Word (read-words Fname) > (if (idx 'Acum Word T) # Find or insert into tree > (inc (car @)) # if found, increment > (set Word 1) ) ) ) # otherwise init to 1 > (mapcar > '((Sym) (cons Sym (val Sym))) > (idx 'Acum) ) ) ) > > Cheers, > - Alex > -- > UNSUBSCRIBE: mailto:[EMAIL PROTECTED] > -- UNSUBSCRIBE: mailto:[EMAIL PROTECTED]
