Hacker News Reader

Comment by rpd9803 | Hacker News Reader

I mean, its a bit apples or oranges.. XML's well-formed rules are, in my experience, fairly universal, hence parsing isn't as much an issue as is poorly constructed XML schemas (and schma-less xml documents as well) I know there are edge cases, but overall XML doesn't have many of the problems specified.

View on HN

XML doesn't have the problems in theory, but in practice it can be just as bad.

As a mildly egregious example, I had a company which wanted a data feed from us in XML format. I wrote a query to produce the relevant data out of MSSQL and sent it to them. They wrote back that I had to have a header line:

    <?xml version="1.0" encoding="UTF-8">

MSSQL writes SQL in UCS-2 format; I therefore sent them back a feed with the encoding declared as such, and they rejected it. They were fine with the encoding being UCS-2, but they were absolutely insistent I had to declare it as UTF-8 or they wouldn't accept it. After a lot of back-and-forth and quoting of standards at them, I finally gave up and wrote a little script to change the declared encoding so they would accept it.

It's this sort of thing that makes me understand why Ode to a Shipping Label[1] exists.

[1]: http://imgur.com/4J7Il0m

A regional (to me) real estate web site that we had to work with had a data service that output XML. It was naively implemented and managed to output malformed XML. Mostly in the form of a CSV field crammed into an XML field with non-conformant XML markup.

Additionally there aren't as many tools that generate XML output as there are for CSV

XML has a specification. There are two well-defined categories of well-formed and ill-formed XML files, and there's exactly one correct way to parse the well-formed ones.

You still write a list like

1: it is well formed

2: the person who produces it will fix it when you tell them

3: the encoding is correct

etc.

Even ycombinator's rss feed is only well formed but not fully specified

http://validator.w3.org/check?verbose=1&uri=https%3A%2F%2Fne...

Reply on news.ycombinator.com