this is on clojure 1.1 (current ubuntu stable rep) so, the code below shows ABS returning 2 different values for the 'same' input in clojure. input being Integer.MIN_VALUE. of course one input is an int and the other is a bigint, but we can't see the difference in clojure, and as the code shows at the end, clojure says they are the same input.
@CuppoJava: to take more care in how we code is a good idea, but i feel that rich hickey doesn't want users of clojure to be thinking about if a certain function is going to act a certain way for inputs that clojure says are the same, but java says are different. i think this type of problem is on the same level as managing the memory of your application. user=> (def min-int-1 (+ -1 min-int)) #'user/min-int-1 user=> min-int-1 -2147483649 user=> (- min-int-1 min-int) -1 user=> (Math/abs (+ 1 min-int-1)) 2147483648 user=> (Math/abs min-int) -2147483648 user=> (= (+ 1 min-int-1) min-int) true @Sean Corfield: I haven't tried this on clojure 1.3. is it part of the clojure road map to eliminate the surprises built into the java language? I think that it would be easy to eliminate a majority of them. a good resource is the book 'java puzzlers' http://www.javapuzzlers.com/ which provides the code in the book for free on their website. not all of the surprises will apply to clojure because they'll be avoided by avoiding the use of java, but some will still apply, and they should be considered in the definition of the language, and when using interop code. so, in summary, ABS in java has a bug/surprise. clojure uses this directly. so clojure has a bug in abs. it's really unlikely that someone will pass abs an Integer.MIN_VALUE, but maybe it should be considered. i think the more likely thing to happen is that abs will be passed a number that is a result of another number being manipulated. so one solution would be to change to a bigint earlier than the language does now. or maybe closure could make a safe abs wrapper. I know it's not a big deal, but when you write a program that uses all integers (quite easy to program in clojure), and also uses abs, and it breaks when it hits this bug... Rich hickey's answer to why he made clojure was that he wanted to write programs without going crazy, fixing some of the stuff that is broken in java, that clojure uses, seems like it follows that spirit. On Nov 2, 9:33 pm, Sean Corfield <seancorfi...@gmail.com> wrote: > On Tue, Nov 2, 2010 at 3:22 PM, box <somethingital...@gmail.com> wrote: > > Math/abs doesn't always return positive numbers. example: > > Have you tried this on 1.3.0-master-SNAPSHOT? > > user=> (def min-int -2147483648) > #'user/min-int > user=> (Math/abs min-int) > 2147483648 > user=> (def min-int+1 (+ 1 min-int)) > #'user/min-int+1 > user=> min-int+1 > -2147483647 > user=> (Math/abs (- min-int+1 1)) > 2147483648 > user=> (Math/abs (- Integer/MIN_VALUE 1)) > 2147483649 > -- > Sean A Corfield -- (904) 302-SEAN > Railo Technologies, Inc. --http://getrailo.com/ > An Architect's View --http://corfield.org/ > > "If you're not annoying somebody, you're not really alive." > -- Margaret Atwood -- 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
