I did a late night coding session the other night (haven’t done one of those in ages) to update the OzPolFeeds – in part to prepare for playing with Atomflow. Lots of internal re-organisation. The service now attempts to handle common parsing errors and now logs errors and continues rather than failing completely. The biggest change, though, is that all of the feeds should now correctly validate (including the Greens feed which uses the most messy and invalid URLs I’ve seen on a site in a long, long time – a lot of tweaking was required to get them working correctly as GUIDs for the feeds without breaking the actual link to the article).
I have also moved the service to a server that is online 24/7 so that it can keep running all the time without interuption. In doing so I established a system that would assign a time to the releases as they were added to the corresponding website. For example, the service polls the sites every hour, and when it detects a new entry dated for the current day, it assigns the current time to the entry. The idea being that as things are added during the day they then correctly appear in order in your favourite news aggregator. Unfortunately it appears that most sites are adding their entries retrospectively – the ALP site, for example, adds entries from Sunday on Monday etc. which means the times are never assigned.
I have a question for the people that use these feeds – is it important that the RSS entry date be the date on the media release? Or would this be better/ok to be the date/time added to the site? The benefit of the latter is that aggregators will display the entry in the order they were added rather than lumping them all together. TIA.