No, Safari is totally broken, as it does what is called "MIME-type sniffing", ignoring the explicitly-declared
text/plainMIME and overriding it due to the presence of markup.
This is actually an excellent example of the dangers explained by the article. What would happen if a user agent "did a Safari" and decided to parse all documents with an XHTML doctype as
application/xhtml+xml? The vast majority would break, even the "valid" ones. Of course, any error in a document served as
application/xhtml+xmlis a fatal error, making the document totally inaccessible.
[edited by: encyclo at 3:50 pm (utc) on April 2, 2006]