Help us solve podcast problems

July 10, 2006 9:32 PM

Podcast headaches!
posted by cortex (staff) to MetaFilter Music at 9:32 PM (8 comments total)

1. The emdash character in this song breaks any playlist feed that contains it, for iTunes at least.

2. The html entity in this song's title line elicts (apparently non-fatal) complaints from Feed Validator.

Someone who knows more about encoding details please hope us. Escaping/nixing these sorts of things properly at feed-generation time seems like the simplest solution, but my opinion is not backed by much understanding.
posted by cortex at 9:34 PM on July 10, 2006

1. In the HTML there's a fully formed UTF-8 emdash followed by a stray 0xC2 byte (no idea why). In the RSS feed the emdash is replaced by a question mark character and the 0xC2 by 0xA0 (presumably you're doing an unnecessary UTF-8 to Latin-1 conversion).

2. That depends what the user entered. If they typed a simple ampersand, then Matt is escaping the RSS title one too many times. It the user typed an entity, the HTML pages should display the whole entity rather than just an ampersand, and the RSS feed is fine.
posted by cillit bang at 11:32 PM on July 10, 2006

I changed the two examples here for now, I hope that doesn't make troubleshooting the problem impossible.
posted by jessamyn (staff) at 6:23 AM on July 11, 2006

Any suggestions for a non-iTunes pod-cast xp app? (for the rss inept)
posted by Hanover Phist at 7:21 AM on July 11, 2006

Sorry, cortex. RSS has historically had a lot of problems with handling HTML entities. Maybe part of the "upload song" script should filter these from titles and descriptions?
posted by Mr. Six at 7:51 AM on July 11, 2006

That, or a filter in the podcast-generation code. Or, realistically, perhaps an email to mathowie/jessamyn every 150 songs or so to request a tweak. Not really sure what would practically be more work in the short term, but in the long run its seems like a programmatical solution would pay off.
posted by cortex at 7:54 AM on July 11, 2006

Sorry, cortex. RSS has historically had a lot of problems with handling HTML entities. Maybe part of the "upload song" script should filter these from titles and descriptions?

That's not really true anymore. Now it's just exposing holes in how a lot of websites are put together. See my rant here.
posted by cillit bang at 8:59 AM on July 11, 2006

Good point, well made, cb.

RSS is XML, and XML needs to be valid, not just sort of, pretty much, close-enough-for-government-work, near-as-damn it work in a browser.

I get the feeling that understanding of what "valid" really means is very very limited out there, let alone character encodings.
posted by AmbroseChapel at 4:41 PM on July 11, 2006