If the data fits (comfortably) in memory why not
just parse it to a DOM tree and build a hash tree
over the nodes you want to key off.
j.
Curt Sampson wrote:
>
> On Tue, 26 Feb 2002, Christian Molstrom wrote:
>
> > ...but I wonder what performance difference there might be
> > between (a) pulling the data straight from xml (probably via SAX) and
> > (b) returning data from a getter method in an object.
>
> Massive. In the second case, you've got the data in an object in
> memory, and you're just returning a base type or an object reference.
> In the second you're doing parsing of strings (always expensive),
> perhaps file I/O, and who knows what else. That's really something
> you want to do only once, if you can manage it. In this case we're
> dealing with quite a small amount of data, so it's not a big deal.
> Even if it got as large as 2K/phone and we had 500 phones, that's
> still only a megabyte of data.
>
> > You need to add a step for parsing the incoming user agent for lookup
> > and figuring out just what object to instantiate.
>
> Well, the idea is that the object is already instantiated; you just
> need to determine to which one to return a reference. But that's
> a minor nit.
>
> You're right that dealing with user-agent -> object lookup is going
> to be a bit of work. But hey, start with something basic, and you
> can refine it as you discover problems. As your library of test
> cases grows, your code will get better and better at this.
>
> cjs
> --
> Curt Sampson <cjs_at_cynic.net> +81 90 7737 2974 http://www.netbsd.org
> Don't you know, in this new Dark Age, we're all light. --XTC
>
> This mail was sent to address jon.ellis@eng.sun.com
> Need archives? How to unsubscribe? http://www.appelsiini.net/keitai-l/
Received on Wed Feb 27 04:23:01 2002