Thanks all, that makes a lot of sense. To summarise, if I understand correctly, (on the JVM) comparing with = : - first checks if the objects are one and the same, returns true if so - if not, checks whether they are numbers or persistent collections, which have special-case checks - if neither of the above, delegate to the object's .equals method. Keywords, as best I can tell, don't override Object/equals, so the test is the same as identical? in the end i.e. whether they are the same object.
It makes sense, then, that comparing keywords that are the same is fast (~4ns, same as an identical? comparison on my machine), as this returns on the first step above. And comparison of differing keywords is slower as it goes through all of the steps above, as noted by @Mikera. I also notice that the code for core/case seems to suggest that if the cases are labelled by keyword, then an "identical" check will be used (although, I don't yet understand the compiler code well enough to verify that this is true in the implementation of case*). I guess a question that remains is whether this behaviour of keywords, that identical? and = give the same result, is something that is guaranteed, or whether it should be viewed as an implementation detail? If it is a guarantee I wonder whether there should be something on the data structures page of clojure.org indicating this, and noting how to do fast keyword comparison? Another question, which is more for my understanding, is given that is currently the case that identical? and = are equivalent for keywords (and some core code depends on this behaviour), why does it work? Looking at Keyword.java I see that weak references are stored to interned keywords in a table. But if any of these Keyword objects were garbage-collected, I think it would break the parity between identical? and =. So it must be the case that something else is holding on to strong references somewhere. Which makes me wonder why weak references are used in the table. Anyone familiar with the compiler able to shed light on this for my edification? :-) Jony -- 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 --- You received this message because you are subscribed to the Google Groups "Clojure" group. To unsubscribe from this group and stop receiving emails from it, send an email to clojure+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
