Sally Justice had already installed version 1.1 of the Expiry Checker and put the script to invoke it into a couple of her web pages. As far as she was concerned everything was working fine, until she received an email yesterday afternoon from one of her colleagues to say that he was getting an error message when viewing one of her pages. He was using Netscape 4.7 on a PC running Windows NT. I explained to Sally that I had experienced a similar problem when using a Macintosh, and I showed her the fix that I'd incorporated into Version 1.2 yesterday morning.
As I had suggested, she took out the carriage-return character immediately prior to the "?" - but her colleague was still getting an error message when he tried to view the page! Meanwhile, I also discovered that although this prevented the problem occurring on a Macintosh, it now wasn't calling the Expiry Checker correctly. The query string was wrong - I think because it incorporated the carriage-returns that were embedded within the HTML source. That didn't seem to be happening when viewing pages on a PC!
Coincidentally, I had briefly discussed the Expiry Checker with Fintan Culwin in the morning and he queried whether there should be " " marks within the script section that invokes the Expiry Checker. I mentioned this to Sally. When she edited these out her colleague didn't get an error message any more when viewing the page, *but* after further experimentation we discovered that the Expiry Checker now wasn't being invoked at all!
This was becoming very frustrating. I decided to create a set of alternative versions to try and find out once and for all what worked and what didn't work. Last night I put together the following >>8 test pages (edited here so that the source HTML is dispalyed), and I invoked each one from a Mac. I discovered that only the last four generated email messages. Not only that, but I was working over a modem and I realised that the time the pages were taking to load varied considerably. The first six took quite a long time because (according to the browser messages) the Expiry Checker script had to be invoked before the page could be displayed. By contrast, the last two loaded much more quickly. This made me realise that using a hidden image to invoke the Expiry Checker was actually a better way of doing it.
The last thing I did last night was to email Sally's colleague asking him to please check out each of the eight test pages and let me know which ones give rise to error messages. I'll also then see which ones send me email messages. (The date will have changed by then, so it won't matter that I've already invoked the test pages.) It's now Friday morning and I'm waiting for Sally's colleague to email me to tell me how it went.