Greg Murray at the University of Virginia Library explained in an email to Kevin on 2008-12-02 that he invented the element name <cols/> when setting up Virginia's vendor specs but has regretted the name since. It would be more consistent with other TEI element names to use <colShift cols="___"/> instead of <cols n="___"/> because:
a) it is consistent with other TEI element names like <shift/> and <handShift/>
b) it avoids confusion with the name of the cols= attribute already used for <table>
c) it better indicates the purpose of the element--to mark a change in columnar layout within a div.
Implemented at https://sourceforge.net/p/tei/code/13088/ . Will email Sebastian to ask him to update tei-xsl package accordingly.
duly made additions to XSL. seems like a weird change to me, as it means vendors retooling for no advantage, when the notation isn't going to persist beyond acquisition anyway. but what the heck.
Sebastian pointed out in a private email that I actually only renamed the element
cols
tocolShift
but didn't also change practice from using@n
to@cols
on this element. I find either attribute name equally transparent, and since I've just added bytes to the name of this element, I am reluctant to also add bytes on the attribute value. So I'll leave it for now.I beg to disagree: the attribute @n should be used (like xml:id) to identify the shift itself, not to describe it. And since the shift being encoded here is not actually in the column (for which <cb/> already exists) but rather in the layout, I think a more appropriate name would be something like <layoutShift> or a contraction thereof if you really care about the bytes. But (as Sebastian says) is it really worth revising this now?
Lou makes a good point about TEI's semantics for
@n
. However, the shift being encoded here is, in fact, in the columnar layout, so I think<colShift>
is more transparent than<layoutShift>
. If we were going to have more than one type of layout shift, we might need<layoutShift>
, but we're not to that point.Both Sebastian and Lou asked whether it made sense to even rename the element when Tite is only meant as a transitory format. It was easy to make the change, and I'd like for Tite elements to be named in a way that would allow them to be added to P5 as syntactic sugar someday in case people wanted them. For that reason,
<colShift/>
is better than<cols>
.Since Lou makes a good case for renaming @cols to @n, I have done so at https://sourceforge.net/p/tei/code/13094/ . Added bonus: this new attribute has the datatype data.count, which will require a non-negative integer -- something that was intended anyway when riding on the global @n attribute.