The ‘feed://’ Protocol
Posted November 26th, 2005 @ 04:05pm by Erik J. Barzeski
When a user encounters a 404 on this site (and a few others I run), I run MovableType's search and show the results to the user (or forward them to the appropriate entry if there's only one match). I also send an email to myself, and lately they've been illustrated that a good many users have problems with the "feed://" (pseudo-)protocol.
Queried URL: http://thesandtrap.com/feed://thesandtrap.com/podcast.xml
Queried URL: http://nslog.com/feed://nslog.com/index.rss
For now, I've taken care of it as such:
RedirectMatch /feed://thesandtrap.com/(.*) http://thesandtrap.com/$1
RedirectMatch /feed://nslog.com/(.*) http://nslog.com/$1
In other words, a lot of people are trying to subscribe to (or at least click on) my feeds but using a browser that both doesn't understand "feed://" and won't simply replace it with "http://" when it fails to understand it.
Donate Life
Posted 26 Nov 2005 at 4:43pm #
Can you tell from your logs which browsers get it wrong?
Posted 26 Nov 2005 at 5:24pm #
Using "feed://" was a bad idea. It gives you nothing and breaks many different things. The protocol being used is http, there's no point in calling it something else.
Posted 26 Nov 2005 at 5:52pm #
Dustin, I disagree. "feed://" was effectively a grassroots effort to get browsers to recognize that the URL should be handed off to something that registered the protocol handler. On the Mac, that's largely true and it's been pretty successful: Firefox, OmniWeb, Camino, and Safari all support "feed://" and applications can register with the system as being the handler for the "protocol."
Jon, the 404 search email I get isn't a log per se, but I can add a line of PHP to the 404 search script so that will tell me. I'll do that now…
Posted 26 Nov 2005 at 6:46pm #
When you say "a good many", what kind of percentage are you talking about?
Posted 26 Nov 2005 at 7:19pm #
Mike, that's impossible to tell since the 404 search isn't triggered when someone's browser properly interprets "feed://".
Posted 26 Nov 2005 at 10:38pm #
Why is a different protocol name necessary to replicate what MIME types do in the http protocol? When I load an RSS feed in Safari in will automatically detect that it's a feed and display it as such, regardless of whether it's linked as http: or feed:. Clearly the feed: isn't needed to detect it as a feed. So the requirement to use the feed: protocol is not a requirement. So what's the point? I see none.
Posted 27 Nov 2005 at 8:21am #
Dustin, it's still comes in handy for those on 10.3 (older version of Safari) and it's still a requirement for those using browsers not named "Safari."
And don't put words in my mouth - you're the first to use the word "requirement." Nobody else has said that.
Posted 28 Nov 2005 at 12:08am #
I think I would rather a link be set up for browsers that are not rss feed savvy, allowing that manipulation to be handled manually. I use Safari, and like its RSS implementation, but really don't want OmniWeb to be automatically fed/ subscribe to a feed (I do not like the way it handles feeds). I haven't even run across a problem with Firefox, which I will use occasionally to read blogs made with blogger, as it seems to parse them with the least amount of effort, but otherwise, firefox is really not doing a lot of RSS or Atom by my intent.
On my desktop I use primarily OmniWeb and Firefox with Pulp Fiction; on my Powerbook I use Safari only.