Well, I escaped Disneyland with my sanity intact. I spent this weekend aimlessly wandering Paris. What an amazing city. I can't believe I've left it this long to visit. I plan on returning with Karen soon and more of a plan on how to spend my time here.

Today I am in La Défense, which seems a bit like London's Canary Wharf. I'm spending a couple of days working on site for Saint-Gobain, a long-time client of mine. It seemed rude to come to Paris and not call in.

First task today was to fix the RSS feed I added to a database of theirs. This really had me stumped for a while. I'll share with you in case it helps you out one day.

The feed is an Agent that churns out XML for documents in the RSS format. This worked fine until special characters were included. Me being English, I foolishly neglected to test it with things like accented characters. Needless to say, them being French, they found the bug straight away. The bug was that IE would not render the XML if special characters were used. It said "An invalid character was found in text content.".

The confusing part of it was that the XML for a view containing the same information as the feed would display fine when opened with a ?ReadViewEntries URL command. My task was to find out the difference and, hence, the reason why.

First thing I did was save the source to both pages - the one that worked and the one that didn't. I then used TextPad to find the difference. After messing about for ages I realised there was no difference in the text and so took a look at the document properties (Alt + Enter in Textpad). Herein lied the problem. The file saved from my Agent output had a codeset of ANSI and a UNIX file-type. The file saved from Domino's XML had UTF-8 and PC respectively. My code was using the wrong character encoding!

Here's the first two lines of code from the Agent:

Print {Content-Type:text/xml}
Print {<?xml version="1.0" encoding="UTF-8" ?>}

The first line tells the browser to expect XML, but not what character set to expect. This is why it complains about anything other than standard text output. To fix the problem I made sure the agent told the browser to expect the same charset as the XML used by changing the first line, like so:

Print {Content-Type:text/xml;charset=utf-8}
Print {<?xml version="1.0" encoding="UTF-8" ?>}

Moral of the story: If you're going to output XML from an Agent, make sure it outputs the text in the same encoding as specified in the XML tag. And, of course, if you're client is Français, soyez sûr d'examiner.