Tomasz Wegrzanowski wrote:
It would be better if numeric entities were converted too. The code is somewhere in Phase1->Phase2 conversion script, and in konwert program too.
I'm not sure what other peoples' feelings on this are, but I personally would prefer using numeric entities on the English wikipedia for characters that aren't standard in English. Many of us don't have keyboard configurations that allow the literal characters to be typed, so keeping them as numeric entities makes editing them easier.
As for the UTF-8 switch itself, I can see some advantages, but I think you gloss over the disadvantages too quickly. On Mozilla Firebird on Linux, at least, the UTF-8 wikis look _much_ worse than the ISO-8859-x wikis, because they're forced to use a Unicode font, of which only fairly ugly ones with strange proportioning are Freely available.
-Mark