oddbyexample.xsl should look at the start element and start with that. I ran over a corpus of files that all were fragments starting with tei:person but yet the resulting ODD has a schemaSpec/@start='TEI'.
I am not convinced that this change would actually be a good idea. If the corpus analysed contains only <person> elements, you might want your schema actually to use <listPerson> as its root. Or <TEI> as per usual. Changing the root element is in any case a trivial change to an ODD. So I would say leave this particular mutt to snore undisturbed.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I agree this needs doing, but its not entirely simple for me, so if you need it soon you'll
need to look at the code yourself I am afraid.
I am not convinced that this change would actually be a good idea. If the corpus analysed contains only <person> elements, you might want your schema actually to use <listPerson> as its root. Or <TEI> as per usual. Changing the root element is in any case a trivial change to an ODD. So I would say leave this particular mutt to snore undisturbed.
Sorry, I am going to reject this, for the reasons Lou gives. Second-guessing user intentions can go too far.
of course, anyone can dive in and add the feature as an option.