It's 2008 and Firefox still doesn't support the feed:// pseudo-protocol. I'm sure there was a time when this was supported.
1 year, 1 month ago.
4 comments so far
I have an utter loathing of pseudo-protocols, and am very glad Firefox doesn't pretend that an HTTP resource is anything else. If you want to handle RSS and Atom feeds differently than HTML or generic XML content, serve them with the correct MIME type. :)
So if I've got a desktop feed reader installed then it will work but if I don't use a desktop feed reader then it's just broken. They should either ignore the pseudo-protocol completely, thus forcing people to stop using it, or use the rules in the spec to take me to the correct http(s) url. The current approach gives me the worst of both worlds.
The problem I have is that if I'm using, say, Safari and I click on a link to an XML feed, it suddenly changes it to feed:// - meaning that if I want to copy and paste the URI of the resource, I have to copy the URI into a text editor, delete the word 'feed' and put the 'http' back in, then copy it again. The way I figure it is that if I can't copy a URI from my browser and paste it as an argument into curl, it's not really a useful URI at all. My browser should tell me the location of the resource in the location bar - not a pseudo-protocol prefix.
The solution was to not jump as quickly as the RSS people tried to - contacted the browser manufacurers and tried to explain to them that they should pivot on MIME type plus content inspection - if it gets back something that smells like XML, have a look inside and if it looks like an RSS/Atom feed (ie. it has the Atom xmlns declaration or the typical elements of an RSS/Atom feed), offer the user the ability to subscribe to it in their favourite application. MIME type plus content inspection seems to solve the problem.
Plus, some people consider having the XML 'spew' into the browser a fault. I consider it a feature. The behaviour of Firefox is a lot more sensible when it encounters XML than Safari - Safari takes all the strings out of the document and just throws them up on the page, whereas Firefox puts the elements and attributes there with syntax highlighting. Of course, this is why you should put some kind of simple XSLT on each XML file you publish, so browsers can present it at least readably.
Of course, eventually, feeds will just be embedded inside the (X)HTML or inferred from the logical structure of the page, and the very concept of publishing a separate feed will disappear. Let's hope so, anyway.
Personally I'm hoping for a future where the idea of millions of clients polling millions of servers only to find that nothing had changed is considered baroque bordering on quaint.
In my preferred future each blog will expose an XMPP or AMQP endpoint. Each user will just register a subscription with each server and when a new post is written it will be pushed to each client that has registered an interest. Users won't know or care about the data formats. There will just be a big obvious "subscribe" button on every webpage. That button will just magically do the right thing in much the same way that dialling someone's phone number just does the right thing.
4 comments so far
I have an utter loathing of pseudo-protocols, and am very glad Firefox doesn't pretend that an HTTP resource is anything else. If you want to handle RSS and Atom feeds differently than HTML or generic XML content, serve them with the correct MIME type. :)
I like the javascript: pseudo-protocol even less.
1 year, 1 month ago by tommorris
I would agree with you but Firefox provides just enough support to ensure that the broken blogs never get fixed. See:
http://www.rassoc.com/gregr/weblog/2003/09/21/rss-and-mime-types/ which lists reasons why mime-types don't quite solve this problem for most people.
http://nick.typepad.com/blog/2006/05/revisitingthe.html
http://en.wikipedia.org/wiki/Feed:URIscheme
http://weblogs.mozillazine.org/ben/archives/015911.html
So if I've got a desktop feed reader installed then it will work but if I don't use a desktop feed reader then it's just broken. They should either ignore the pseudo-protocol completely, thus forcing people to stop using it, or use the rules in the spec to take me to the correct http(s) url. The current approach gives me the worst of both worlds.
1 year, 1 month ago by adewale
The problem I have is that if I'm using, say, Safari and I click on a link to an XML feed, it suddenly changes it to feed:// - meaning that if I want to copy and paste the URI of the resource, I have to copy the URI into a text editor, delete the word 'feed' and put the 'http' back in, then copy it again. The way I figure it is that if I can't copy a URI from my browser and paste it as an argument into curl, it's not really a useful URI at all. My browser should tell me the location of the resource in the location bar - not a pseudo-protocol prefix.
The solution was to not jump as quickly as the RSS people tried to - contacted the browser manufacurers and tried to explain to them that they should pivot on MIME type plus content inspection - if it gets back something that smells like XML, have a look inside and if it looks like an RSS/Atom feed (ie. it has the Atom xmlns declaration or the typical elements of an RSS/Atom feed), offer the user the ability to subscribe to it in their favourite application. MIME type plus content inspection seems to solve the problem.
Plus, some people consider having the XML 'spew' into the browser a fault. I consider it a feature. The behaviour of Firefox is a lot more sensible when it encounters XML than Safari - Safari takes all the strings out of the document and just throws them up on the page, whereas Firefox puts the elements and attributes there with syntax highlighting. Of course, this is why you should put some kind of simple XSLT on each XML file you publish, so browsers can present it at least readably.
Of course, eventually, feeds will just be embedded inside the (X)HTML or inferred from the logical structure of the page, and the very concept of publishing a separate feed will disappear. Let's hope so, anyway.
1 year, 1 month ago by tommorris
Personally I'm hoping for a future where the idea of millions of clients polling millions of servers only to find that nothing had changed is considered baroque bordering on quaint.
In my preferred future each blog will expose an XMPP or AMQP endpoint. Each user will just register a subscription with each server and when a new post is written it will be pushed to each client that has registered an interest. Users won't know or care about the data formats. There will just be a big obvious "subscribe" button on every webpage. That button will just magically do the right thing in much the same way that dialling someone's phone number just does the right thing.
1 year, 1 month ago by adewale