On Thu, Jun 23, 2011 at 18:48, Ken Wesson <kwess...@gmail.com> wrote: > On Thu, Jun 23, 2011 at 11:55 AM, B Smith-Mannschott > <bsmith.o...@gmail.com> wrote: >> user> (name 'a/b/c) >> "c" >> user> (namespace 'a/b/c) >> "a/b" >> >> Is this intentional? I would have expected a/b/c to be >> rejected as a symbol name since we use slashes to separate >> namespace from name and conventionally use "." to indicate >> hierarchy in namespace names. > > => (namespace (symbol "" "a/b")) > "" > => (name (symbol "" "a/b")) > "a/b" > > Both the namespace and the name can contain slashes, but the normal > reader behavior is that foo/bar/baz is interpreted as namespace up to > the last slash and then name. Unless the symbol is just "/", in which > case the reader treats that as the name, with no namespace. (Necessary > for a bare / to resolve, ordinarily, to the division function > clojure.core//). You can get any desired symbol using the two-argument > form of the symbol function, though. > > There's some ickiness here. For instance the one-argument form of > symbol doesn't parse the same as the reader does: > > => (clojure.core// 42 2) > 21 > => (eval (list (symbol "clojure.core//") 42 2)) > #<CompilerException java.lang.StringIndexOutOfBoundsException: String > index out of range: 0 (NO_SOURCE_FILE:1)> > => (eval (list (symbol "clojure.core" "/") 42 2)) > 21 > > In fact, (symbol "clojure.core//") produces a symbol with namespace > "clojure.core/" and name "", the latter of which apparently leads to > the rather obfuscatory error message above. > > Symbols whose names contain slashes don't work as vars -- another > cryptic message, since the symbol's actually unqualified here: > > => (eval (list 'def (symbol "" "foo/bar") 42)) > #<CompilerException java.lang.Exception: Can't refer to qualified var > that doesn't exist (NO_SOURCE_FILE:1)> > > And symbols whose namespaces contain slashes won't work as vars with > AOT, if at all, since the namespace name ends up a class file name and > slashes are path separator characters on every sane filesystem and > thus cannot appear in file names. > > Moral of the story: try to avoid slashes in symbol names, except for > the unqualified name of the division operator. :) But if you really > for some reason need one it can be done, though the only apparent use > for such a symbol is as a kind of auto-interning string rather than as > a var name. You'd probably be better off using a keyword for such > uses.
Thanks! My question probably should have been: is it intentional that the Clojure reader accepts symbol names containing more than one slash, producing a namespace portion of the symbol containing slashes in its name? I've been reading through LispReader.java and porting bits of it to Clojure TDD-style as a finger exercise. That's how such questions arise. // ben -- 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
