Help Me Not Kill The Feeds

by Erica Sadun

Late last week, Mac DevCenter Blog's feeds went a little screwy. Mea culpa. I used an uppercase closing tag to match a lowercase opening tag.

Some feeds (in particular Atom 1.0) depend on valid xhtml so news readers will parse them correctly. Unfortunately MT does have any kind of HTML fixer/checker. So I ask our readers will the OS X command line xmllint --html work properly and consistently as a validator to help this problem?


Gordon Meyer
2006-03-13 13:55:12
I think '%tidy -e' is more what you're looking for.
Erica Sadun
2006-03-13 14:51:08
Fantastic! Thanks for the pointer.
2006-03-13 22:12:53
Atom doesn't strictly require valid XHTML. It supports three flavors of text: plain, valid XHTML, and escaped HTML. In the latter, any kind of nasty tag-soup is acceptable (since it's all just a bunch of text as far as the XML parser is concerned).

If you end up being unable to make your CMS generate correct XHTML, you'd better switch your Atom feed generator to use "html" mode.

Aristotle Pagaltzis
2006-03-14 17:19:55
I second Jens’ recommendation. I wrote this to the webmaster team at O’Reilly on multiple occasions on which the Atom feed broke back when the new feed infrastructure was being phased in. If you can’t guarantee well-formedness, then you should transport it in type="html".