(keitai-l) Re: open source keitai tools (was Re: Re: western phone, imode si

From: Curt Sampson <cjs_at_cynic.net>
Date: 02/26/02
Message-ID: <Pine.NEB.4.43.0202261141210.413-100000@angelic.cynic.net>
On Tue, 26 Feb 2002, Jon Ellis wrote:

> http://www.ccpp.org/ might be worth looking at...

I just downloaded it and checked it out. A few random comments.

1. These guys really don't believe in documentation (beyond "see
the website", which says only "coming soon").

2. They use XML files to describe the capabilities of the devicesk,
which is good.

3. The guy just *loves* String objects. Not only does he create
(possibly large) Strings when he should just be passing around a
Reader-type object (a common mistake in Java that destroys
performance), but at one point, he passes a list of items by
constructing a String containing comma-separated names, which he
then parses! I guess he figured that if it worked in perl....

4. It doesn't feel terribly object oriented. When you start making
classes with static methods, and passing objects into these methods
to get information out of them, you can smell that something's not
quite right there. I think in part this is because it's based on
DELI, which also does not feel terribly OO. (Things like returning
a String when you get a resolution rule--how on earth do you parse
it and use it?)

5. It's based on DELI, for which you can get more info here:

    http://www.hpl.hp.com/techreports/2001/HPL2001-260.html
    http://www-uk.hpl.hp.com/people/marbut/deli/

DELI, which I also downloaded and looked at, seems from the paper
in the first link above to be a lot more complexity than the problem
really justifies. Do we really need a huge pile of default resolution
rules to determine whether or not a device has sound? It seems to
me to be easier just to say "yes," "no," or "the profile didn't
say." And that avoids getting a default value from somewhere that
is in fact incorrect; I should think most applications would rather
know that they don't know than use a wrong value.

I also sense a sort of basic problem in DELI: it's supposed to let
devices tell you about their capabilities, so you can change things
to suit them. But one of the main reasons you want to set up a
device database like this is due to bugs in devices; they don't do
what they're supposed to. That probably means they also don't do
what the profile says. Now DELI does seem to offer the ability to
override some of the information you get from the device by
downloading, say, an bug report from the manufacturer's web site,
but the complexity here is starting to get quite disturbing.

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
Received on Tue Feb 26 05:30:34 2002