A bit of feed confusion here:
- This validator says my RSS is valid.
This validator says my RSS is invalid.
Does not compute. Does not compute.
Update: Hooray! As mentioned in a response from Andrew Grumet, my feed now validates. Thank you!
-
About Me
Hi there! My name is Leslie Michael Orchard. I'm a serially enthusiastic, caffeine-dependent {web,mad,computer} scientist and {tech,scifi} writer working for the Mozilla Corporation and living near Ann Arbor / Detroit in Michigan.
-
Buy my books
16 Comments
Now you’re just being silly. Of course the Userland validator doesn’t validate RSS 1.0
It used to. The scripting.com validator is just a fork of our validator, and they gave identical results on all input when Dave forked our validator 3 months ago. This difference is new since then, and we haven’t touched the RSS 1.0 validation rules.
Rogers Cadenhead has put up a test RSS 2.0 feed which exhibits similar behavior:
http://www.cadenhead.org/hype/index.xml
In the interest of debugging, I borrowed a copy of Rogers’ feed and shuffled the order of things around, so that dc wasn’t the first namespace declared, and wasn’t the first namespaced element encountered. The dc elements remain the only ones reported as errors.
I wish I knew the code better, to know exactly what it would take to have that happen to it.
Or maybe I do - looks to me like the only way you would get that particular error would be if src/feedvalidator/base.py didn’t have the line for dc in the namespaces hash.
Les, are you doing the Time Warp, again? Looks like comment timestamps are something like 14 hours slow.
Phil, if there were a version of the feedvalidator without that line present, the net effect would be that the dublin core namespace would be treated as “unknown” and you would see results very different from you are seeing: namely that everything in that namespace would be treated as a valid extension.
I’ve sent a note off to Andrew Grummet asking if he knows of any changes to the rss.scripting.com web site that would explain this.
Funky.
Working on it…
I think this is fixed now. If you happen to know when the behavior change occurred, could you send me a note at aegrumet@alum.mit.edu? Thanks.
0xDECAFBAD dude, I think your CSS has the IE6 peekaboo bug (do a Web search on it). I fixed this on my own weblog by adding a “line-height: 1.3em;” to the .content section of my stylesheet.
Thanks for the heads up, and Phil, thanks for the debug suggestions. I don’t yet understand what went wrong but we didn’t fork any of the project files and the fix didn’t require file edits; it was tweak in our setup. After putting the baby to bed I’m going to dig around a bit more.
Andrew: Yay! Makes sense now…
And, ugh, I was was wondering what the hell that bug was, and if it name. Grr.
Andrew Bowden [1] might know, since he says he was signed up for validation reports by email and the errors cropped up “last week.”
[1] http://www.bods.me.uk/rantnrave/2004/04/03/new_atom_full_post_feed.live
I don’t think this is going to be my week. Can’t read code, can’t follow simple directions.
Er, “Hey, Les! Could you email Andrew and tell him…” Sigh.
But at least I screwed up in a new way: I think that’s the first time I’ve carefully checked whether HTML was allowed, noticed from the preview that it wasn’t, copied in an MT-produced URL, and had the filename Textiled into oblivion :)
Excellent detective work, Phil.
Here’s a link that should work: http://www.bods.me.uk/rantnrave/2004/04/03/new%5Fatom%5Ffull%5Fpost%5Ffeed.live
One Trackback
Mystery Solved
Andrew Grumet and I finished debugging this problem yesterday. It took a number of sessions, via AIM, phone, and email, to finally track it down.