* Bruno Le Floch <[log in to unmask]> [2012-02-03 15:01:01 -0500]: : Joseph is right. Regular expression matching is already tough enough : to not try to do it expandably (besides, that would become : unreasonably slow, and would break in corner cases). What you seem to : be trying can be done as follows instead: : : \regex_match:nnTF {o} {o} : { \xdef\foo{0,0} } : { \xdef\foo{1,1} } Thanks Bruno. I finally manged to do the parsing correctly. I had to jump through lots of hoops. I have a couple of comments about the regex API (overall it's fine). o On pages 2--4 I'd give some examples for each of the different regular expressions. I've use regular sed expressions a lot but I couldn't always understand the perl/POSIX regexpression. In the following, I've added a few questions about this. o Suggestion: Instead of Quantification, you could consider closure (as in .*: Kleene closure). o What is the difference between greedy and lazy matching? o I think I understand the difference, but it would be nice to explain the difference between capturing and non-capturing groups. o In the documentation, it is not clear what you mean by <token list>: is it a TeX token list or an expl3 token list variable. o I really missed a regex_match for matching on an expl3 token list argument:) o Page 7: zeroeth -> zeroth. Regards, Marc van Dongen