On Fri, Dec 10, 2010 at 12:40 PM, Thomas Schmitz
processing an attribute with the 'xml:' namespace has been fixed by Hans very recently (last weekend), so if you haven't updated to the latest beta version, please do and try again; your syntax \xmlsetsetup{#1}{div[@xml:id='a']} is basically right.
I've done as you suggest and downloaded the minimals beta leaving the TeX-live distribution behind, so I can be on the same page with you all. Hans's fix has worked. I'm gratified that my intuition about how @xml:id should work was not wrong. The example in my last message now works as expected.
Addressing content that has been set in xml anchors is possible, but not quite easy. I wouldn't have been able to figure it out without Hans. Just to show you an example: here's code that in my TEI document will fetch the content of a witness list:
\xmlfirst{#1}{root::/TEI/text/front/div[@type=='sigla']/listWit/witness[@xml:id==string.sub('\xmlatt{#1}{wit}',2)]}
This looks sensible to me. I'll see if I can't make it work with my document next week. Thanks ever so much. Is that string substitution function a bit of Lua?
Last point: I don't want to sound negative, but what you're attempting is not straightforward and certainly not easy stuff. I am currently planning on doing a long-term project on editing texts in this manner, and I'm quite optimistic that all problems can eventually be solved. But if you need short-term solutions, a perfectly typeset edition within a matter of weeks, this may not be possible right away. E.g., I don't think we really have a solid framework yet for a bilingual edition as you envisage it. It is possible, but not easy, Hans needs the time and many test cases to make the code work. There is a research group here in Germany (with public funding) who are currently working on xml printing of material of exactly this type. I can assure you that for the time being, there's nothing out there which comes even close to ConTeXt in this area, but we're still at the beginning. So: if you want to help and test and be part of this, you are more than welcome, but none of us has a magic bullet; this is work in progress, not drop-in solutions.
Thomas, I do not take it in a negative way; simply practical. I take seriously both the principle and practice of open-source endeavors like ConTeXt, and the scholarly work it can support. I hope and intend that as I get up to speed, I'll be able to make some small contribution, even if it's only posting the commented example of our project on the wiki. I've done some programming in Python, and am rather keen to learn something about Lua, as they have quite a bit in common. It's unlikely I'll be able to make substantive contributions to the code, but not impossible in the fullness of time. I'm mindful of the generosity of all involved, and all on this list, and offer great thanks and kudos. Your cautionary note on the bilingual edition is taken to heart, and I have recommended that for the time being our project will not try to implement a facing page translation. This remains an important desideratum for us however, and I will attend with great interest your progress, and the insights of any engaged in similar efforts. The specific use-case we have in mind, and the problems associated with it may be of interest to others and I'll post it on the wiki.