Laurent PETIT <laurent.pe...@gmail.com> writes: >> The same logic could suggest we remove "or" because we can express it >> with "and" and "not". > > Except nobody complains about "or", "and" or "not" ;-)
Actually, they are a right pain when writing my library, because I wanted to use them to mean something else. some was irritating for the same reason. >> And cause frustration for people who find typing >> >> (:require clojure.test :refer :all) >> >> when they used to type >> >> (:use clojure.test) > > Code is read more often than written. > Clojure makes default choices "easy", and non default choices "harder" > for a reason: to guide people. And again, I an struggling to see when being able to write (deftest hello (is true)) is a bad thing. I agree, using many namespaces is problematic, I do often use one. > > It's like the mantra "if you find it hard to write, you may be doing > it wrong (though it's still possible to do)". > > Also, for the REPL, there will still be the (use 'clojure.tests) call > that you can use in your REPL bootstrapping code. > >> To me, the discussion seems to be confused; I understand why making an >> implementation simpler is important. But removing a simple declaration >> to replace it with a more complex one doesn't seem to make things >> simpler to me. > > I don't think I'm confused, AFAIC: I'm not even thinking from the > implementation perspective, but from the consumer perspective. > It's also psychological: if you remove :use, even at the cost of > keeping :refer :all, that's one less top-level 'ns directive to > remember. I said the discussion was confused not you! So far, I have see three main reasons for removing use. 1) It offends my idea of truth and beauty: only clojure.core should be unnamespaced. Counter argument 1: using deftest makes life a lot easier. And I have files which have 100s of calls to a one namespace and none to clojure.core. Counter argument 2: the functionality will remain anyway. Counter argument 3: so, don't use it then. 2) ns is too complex Counter Argument: Yes, it is, and it does confuse newcomers. I got confused, not least because it accepts both vectors and lists (so two syntaxes with no difference AFAICT). So (ns a [:use b]) (ns a (:use b)) are the same. It also has short cuts: so (:use [a.b.c]) is the same as (:use [a.b c]), (use library "a.b.c") (:use [a.b c]) is different from (:use [a b c]) (first is "use a.b.c" second is "use a.b and a.c"). It's got :as which means you need nesting. and you can combine this with aliases. And I think you can drop the brackets so: (ns a (:use [b])) is the same as (ns a (:use b)) All of this remains. It's still going to be complex. Don't quote me on any of the statements made here; I was trying to do it from memory to see if I remembered correctly. 3) Removing use will make things simpler Counter Argument: Only for people who don't use it, and they don't use it so who cares? For people who do use it, we now have to do (:require x :refer :all) instead of the more simple (:use x). And does refer all support exclude? So can I do (:require x :refer :all :exclude [a b c])? > I agree that it is micro-optimization, but this counts too, noticeably > because it's one of the first things a newcomer is confronted to when > he starts seriously with the language. I agree that it's good to simplify things. Confused as to how this helps. Phil -- -- 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/groups/opt_out.
