What is Pub Sub Hubbub? Well, apart from being a badly named service, it is basically a hub that sits between a server (publisher) and it’s feed users (subscribers). The server talks to Pub Sub Hubbub when content is updated and the Pub Sub Hubbub server then pushes content to individual subscribers dependent on their preferences.

This is innovative in that it actually pushes the content across the web rather than users constantly polling the service. This means that the publisher can save an awful lot of bandwidth and subscribers can receive the updates practically instantaneously.

You can find out more on the project by workers at Google (working outside of Google) here: http://code.google.com/p/pubsubhubbub/

If you’re using Feedburner to manage your RSS feed subscribers (and lets face it why wouldn’t you be? – note to self….), then you can enable Pub Sub Hubbub support. Click on the feed you want to enable, and under the publishing options, look for “ping shot”, and activate the service.

I mentioned this over at webdeveloper.com a while ago, and it’s still an issue that I’m running in to.

Apache error_logs are showing quite a lot of “file does not exist” errors for pages using rewrites. For example, you could visit “http://www.mydomain.co.uk/product/103697.html” and the error log would show “http://www.mydomain.co.uk/product” with the file does not exist error.

Fine (I thought) as the referrer is the page, it must be that there is a broken link.

No such luck – no broken links on the page, and seemingly nothing wrong. In fact, while I’m looking at the specific site that I noticed the issue on first, it appears that I’m not affecting the error_logs at all. None of my browsing is causing errors.

So, I switched to the access_logs and started looking for times & dates of the errors. Turns out, all users were using IE8.

Cutting a long story short, I figured out – by looking at the number of errors that would appear on each page request – that the issue had something to do with Javascript includes on the page.

All the javascript pages were referenced in the header include file as relative i.e <script type=”text/javascript” src=”js/monkey.js”></script> There has never been an issue with this – until now – and specifically with IE8.

I do seem to remember something about the new rendering engine in IE8 parsing the page several times to pull in content I don’t see how a browser rather than an specific action could be causing an Apache error.

To fix the issue, making the javascript files “more” relative cured the problem. Altering the source to “/js/monkey.js” fixed the issue.

To be honest, it’s got me stumped. Can anyone shed any light on this?