#clojure log - Jul 20 2008

The Joy of Clojure
Main Clojure site
Google Group
List of all logged dates

16:06 Chouser: rhickey: do collections cache their hash value?

16:10 anyway, here's my latest favorite: cononical xml hashes use fully-qualified names (http://cars.com/xml/part)

16:11 elements provide (perhaps as meta-data) a mapping of namespace uris to prefixes and/or vice-versa ({:cars :http://cars.com/xml})

16:12 putting that in meta-data might make sense because: it could still be used on the vector format ([:tag {...attrs...} content...])

16:12 also because: considering nodes to be identical when their only difference is their choice of namespace prefixes might be what you usually want

18:06 to make that work, when parsing xml the namespaces and prefix maps would be populated. When filtering and inserting, the APIs would have to allow the user to point at some kind of prefix map.

18:07 In both cases I think it would be easy to allow them to default to the stack of prefix maps formed by whatever you're searching or inserting into. But you'd have to be able to provide override or substitute prefix maps.

18:08 when emitting also, you'd use the prefix maps in the metadata to determine what prefixes to use when making ns declarations.

18:09 only if fully-qualified names were given by the user without any matching prefix map would emit have to actually generate a namespace prefix out of nothing.

18:09 Anyway, that's what I'll be aiming for, to see if it actually would work with real use cases.

20:41 ericthor: rhickey: Trying to test the debugger stuff we discussed the other day..moved the *.clj into the clojure dir. Modified the ant script to package them there, modified doinit to look in clojure/*.clj

20:41 also modified to the compiler to use clojure. as a prefix.

20:42 Is there anything else I need to do to clojure to test this?

Logging service provided by n01se.net