[Openstreetmap-dev] OSM's Schema - moving it forwards.

[Ben Gimpert]

On Tue, Nov 29, 2005 at 11:58:43AM +0100, Immanuel Scholz wrote:
> So what are the benefits of using XML and why don't OSM use some other
> format as CSV instead?
> 
> - XML does the encoding stuff. You don't need to care about sending , ; ?
> \t \n or \0 within a string.

<beating_dead_horse> CSV also does this, as do a zillion other less
verbose, less wanky formats. Dig:

	http://rath.ca/Misc/Perl_CSV/CSV-2.0.html#csv%20specification

> - XML does the validating stuff. The parser is much more robust about
> version mismatches of client and server or broken servers or clients.

This is arguably a bug, not a feature. Validation at the parser level is
bad news for maintaining code changes. Bear witness to the utter horror
that is Java's default serialization support -- the "long
serialVersionUID" drivel. Better to have a simple, explicit explanation
for each property and keep code separate from model. Humans don't (need
to?) read DTD's or Schemas anyway.

> - XML parsers are widely available. Even a brainfuck XML parser is out there.

Yup. And just because there's a wall of books at Borders don't mean its
a good technology. [Insert standard argument on line-on-CV versus
hackably-useful.] Oh and aren't 13 different, conflicting versions of
Xerxes fun...? </beating_dead_horse>

		Ben