Last night I discovered the hard way that flickr uses slightly different GUIDs in its RSS and Atom feeds for the same entries. The format looks something like this in the Atom feed:
tag:flickr.com,2005:/photo/12345678
And in the RSS2 feed:
tag:flickr.com,2004:/photo/12345678
Notice the difference? The photo ID is the same, the tag (a year, probably?) is slightly different.
I converted everyone’s Flickr feeds on whoisi to use RSS2 instead of Atom because the RSS2 feeds contain the thumbnail as part of the <media:thumbnail> element instead of having to make a separate call to the Flickr API. This makes things like preview and adding new sites super-fast and it will probably let me remove a thousand lines of python and JS from the whoisi source code.
But the side effect was that with new IDs every entry looked new. And I suddenly had a few thousand new entries in my unseen page on whoisi. I spent some time last night and this morning cleaning up the database to reflect the proper GUIDs in flickr feeds.
I have pointed out a major slip like that before to them and it was fixed in days, together with a note from one of their coders apologising. I was impressed.
At this point I really don’t want them changing it again. That would cause all kinds of chaos.
Come to think of it, I’m 99% sure that twitter did this at one point and I had to scramble in that case as well.