In the 'Wikipedia:Bug reports' page, it came up that someone's trying to make an article titled "β-cyclodextrine", ie using the HTML character entity for the greek letter beta (β).
Currently the code makes a mess of attempts to do this; it can be sorted out, but the question is how *should* it behave?
Possibility one: Ampersands in general and HTML entities in specific should not work in article titles; links to such pages will be considered invalid. (I believe this is how it worked on the old software.) The user should probably instead title that page "Beta-cyclodextrine".
Possibility two: Ampersands will be escaped; HTML entities will not work as such in article titles, but the name typed as "β-cyclodextrine" would appear as typed, with an ampersand, the word "beta" and a semicolon. The user should probably instead title that page "Beta-cyclodextrine", but might well call a page "HTML entity β".
Possibility three: HTML entities should work in article titles. Thus the name typed as "β-cyclodextrine" will appear in body text and the page title with a beta character. For "HTML entity β" to display as intended the wikipedian must type "HTML entity β".
Possibility four...?
-- brion vibber (brion @ pobox.com)