Re: The stupidest design I ever saw

Marshall Spight wrote:
> Joseph Kesselman wrote:
> > Mikito Harakiri wrote:
> > > <node id=0 parent_id=null label='A'>
> >
> > Whatever that is, it isn't XML. XML attribute values must be quoted.
>
> Exactly. Once you put quotes around the attribute values, the
> design becomes really good, because text based formats are
> flexible and nice. You can just put in whatever you want. That's
> the right way to manage structured data. "Traditional" databases
> are no good, because tables are flat, and the real world is tree
> structured.

Really? Where is the proof that that? I find hierarchies artificial and limiting when trying to model stuff. They are very poor at handling more than about 3 orthogonal categories.

> There are lots of structures that you can't express
> as a single table, like a linked list.

This is wrong, but somebody already pointed that out.

>
> Another advantage of text formats is that anyone can write a parser
> for them. You want as many parsers as possible; that way the
> market can decide on what variations of the format are acceptable.

Comma-seperated value format is far easier to build a parser for than XML. XML is for people who don't "get" relational IMO. XML is neither good for (most) human eyes nor computerization of info because it encourages poor normalization and over-use of hierarchies. If you go down the nested text route, then LISP ess-expressions would be better anyhow. It is usually better repetition factoring, for one.

>
>
> Marshall

-T- Received on Sun Apr 09 2006 - 09:06:44 CEST