RSS Synchronization, Some History
Stefan Tilkov got me thinking about a subject that I used to deal with daily, but haven’t in ages. The main reason for that is that my life no longer revolves around syndication, and to the extent it does, I do everything in Google Reader. But it’s still an interesting problem and got me to thinking about the possibilities. First, some background on what Stefan was talking about:
There’s definitely prior art out there. Dare Obasanjo did some work in this area with SIAM. It seems like the spec is gone, but IIRC it was something like OPML annotated with GUIDs and MD5 checksums of the content. The systems I’ve either looked at or worked on and with personally include:
These systems amount to forcing a download from the central server since they’re implemented in the feed itself, and that’s anathema to Brent’s desired system. At NewsGator, the original idea was to sync between copies of NewsGator Inbox, and the content wasn’t so much the issue, but we hadn’t cracked the problem of positively identifying an item, mainly because RSS GUIDs had been proven unreliable – even if they were present, they were sometimes useless, often enough that we considered them as only partly reliable. Using a secure hash to id items requires a well-known way to generate the hash, including eccentricities like whitespace handling and using a normalized encoding (since every bit counts) which are surmountable, but need to be specified explicitly for a hash generated on one system to be useful on other systems. It also introduces things like the problem of trivial refresh in an item; for example a user’s ill-served by changing ads in a feed showing up as modifications to the feed, or really by something like republishing a feed (which Google Reader still seems to be confused by). On top of all that, there’s the problem with positively identifying a feed: are http://www.example.net/rss.xml and http://example.net/rss.xml the same feed? How about http://foo.example.net/rss.xml ?
Those are just the thoughts that pop to mind. I’m not saying there’s anything preventing a good enough solution, but those are some of the issues out there.
— Gordon Weakliem
Comment
Commenting is closed for this article.
Good Command Line Arguments to Powershell Scripts A Good Start