Hi, Am Dienstag, den 08.04.2008, 11:51 -0700 schrieb Dan Weston:
Paul Johnson wrote:
You can regard an "instance" declaration as an inference rule for the type checker, with "=>" meaning "implies" (though I don't think its the answer to your other question about names).
"implies" might be a bad word, because the direction is backwards:
Eq a => Ord a
is clearly false if read that "Eq a implies Ord a". It is Ord a that implies Eq a.
I just read it as "required for" myself.
I think both can be justified. Take
instance X a => Y a where ...
"X a" is required so that the instance declaration can be used. But also: Given this declaration, having an instance "X a" implies that you will also have an instance "Y a". But not not every type with "Y b" will also have "Y a". Or were you talking about constraints in the class declaration, which your choice of Eq and Ord suggests? Greetings, Joachim -- Joachim "nomeata" Breitner mail: mail@joachim-breitner.de | ICQ# 74513189 | GPG-Key: 4743206C JID: nomeata@joachim-breitner.de | http://www.joachim-breitner.de/ Debian Developer: nomeata@debian.org