Re: MV Keys

From: Jon Heggland <heggland_at_idi.ntnu.no>
Date: Tue, 7 Mar 2006 08:48:36 +0100
Message-ID: <MPG.1e77593ac35bf2a98978f_at_news.ntnu.no>


In article <1141660477.863301.110840_at_i40g2000cwc.googlegroups.com>, marshall.spight_at_gmail.com says...
> Jon Heggland wrote:
> > Huh? The algorithm has to exist somewhere. Is the issue just how easy it
> > is for the user of the library/language to see it?
>
> Agree that the algorithm has to exist somewhere. But the form
> it can take as a library function is limited and probably monomorphic;
> as part of the language, it could be a parameterized code generator;
> it might not look anything like a Java method or a stored procedure.

Ok... but how is this significant?

> > What advantage do Java arrays get from being "in the language" as
> > opposed to the Collections API "in the library"? I know that the
> > collections may not be present in special Java subsets (Micro Edition
> > and such), but is that all? Is it the special syntax that is important?
>
> Syntaxt can be important, but the issue I'm concerned about
> is semantics. In particular, I've seen lots of subtle difficulties
> caused by the fact that types systems generally don't distinguish
> between ordered and unordered data.

Can you give any examples? Is this a problem with Java's Collection, Set and List?

-- 
Jon
Received on Tue Mar 07 2006 - 08:48:36 CET

Original text of this message