the second one is the fast alternative; i always keep the previous versions of implementations for educational purposes; take whatever you like most, i can change them afterwards
Right, that makes sense. I'll use the second one.
Looking at the second vector definition, it starts at char 159 and so looking down to position 176 there appears to be the correct code - namely \textdegree. This presumably explains why it appears correctly in running text in my sample posted to the ConTeXt list. But why does it not work in the XMLdata section? Is it the case that there is another unicode processing system for data being processed as XML? Is this where the fix is required?
hm, aren't the math chars in another vector then?
Most of them are, but the most commonly used (I guess?) are in 0. Any idea what might be creating the $^$ output when the textdegree character is used in XML? Looks like it maybe isn't a unicode problem?
\startunicodevector 34 \unknownchar \or ... [repeat 41 times] ... \cup \or ... \stopunicodevector
one option is to make a full vector (255 codes) with unknowns and to fill then stepwise
Great, understood. I'll get onto this then. Thanks, Duncan