I named is-number-regex poorly. I meant it to be a function that calls re-matches. Here's a more complete snippet, with better names:
; bad code (filter #"^\d+$" maybe-numbers) ; good code (defn re-match? [re s] (not (nil? (re-matches re s)))) (defn number-text? [s] (re-match? #"^\d+$" s)) (filter number-text? maybe-numbers) My point is that, since magic numbers are bad, you should be giving the regex a meaningful name, so implementing IFn for regexps isn't a big savings because it pretty much just transforms: (defn number-text? [s] (re-match? #"^\d+$" s)) into this: (defn number-text? [s] (#"^\d+$" s)) On Aug 27, 12:50 pm, Sean Devlin <francoisdev...@gmail.com> wrote: > Hmmm... I think you're confusing the issue. Your compliant seems to > be more directed at "magic numbers" than regexes. If I understand > your argument (which I agree with): > > ;bad code > (filter #"\d+" maybe-numbers) > > ;good code > (let [is-number-regex #"\d+"] > (filter is-number-regex maybe-numbers)) > > Granted, some understanding of regexes is required. However, if > you're hacking Lisp, I think you've moved beyond beginner status, and > expecting a basic familiarity of a regex is fair. > > My $.02 > > Sean > > On Aug 27, 1:37 pm, eyeris <drewpvo...@gmail.com> wrote: > > > I have the same urge, to want to use regexps as predicates. However I > > definitely would not like to read such code. I can only imagine having > > to try to read such code if I didn't understand regexps. E.g. (filter > > #"\d+" maybe-numbers) is clear enough to someone who understands > > regexps. However (filter is-number-regex maybe-numbers) is clear to > > even programmers that don't understand regexps. > > > On Aug 27, 10:41 am, Sean Devlin <francoisdev...@gmail.com> wrote: > > > > On Aug 27, 1:06 am, Timothy Pratley <timothyprat...@gmail.com> wrote: > > > > > > Granted, this wouldn't work for anything that gets passed to Java, but > > > > > the following gist would be a start. > > > > >http://gist.github.com/176032 > > > > > You already have a getPattern method for those cases. Which suggests > > > > another solution: > > > > > (defn re-fn > > > > "Uses ss to construct a java.util.Pattern. > > > > Returns a function which returns the Pattern if called with no > > > > arguments, > > > > and calls re-seq if called with a string argument." > > > > [ss] > > > > (let [pp (re-pattern ss)] > > > > (fn re > > > > ([] pp) > > > > ([s] (re-seq pp s))))) > > > > > user=> ((re-fn "2.") "12324251") > > > > ("23" "24" "25") > > > > user=> ((re-fn "2.")) > > > > #"2." > > > > > If #"X" created a (re-fn "X") then all the re functions could accept > > > > a function and call it in order to avoid having to do (re-find (pp) > > > > s). The previous signature could be retained so that they would work > > > > with either type of arguments if that were desirable. The downside is > > > > trying to explain that in the docs might be confusing - so a wrapper > > > > seems more obvious for that. > > > > > Oh - it seems like re-seq does the most work so perhaps that is the > > > > best candidate? > > > > The only feature I want is the ability to use a regex as a predicate. > > > So, I'd prefer something like re-matches. Maybe this isn't the > > > biggest use case, though. > > > > Sean > > > > > Regards, > > > > Tim. --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en -~----------~----~----~----~------~----~------~--~---
