As a general remark, indices need not be 2 letter, even though it is recommended to use 2 letter names.
On Tue, Dec 11, 2012 at 07:20:21PM +0100, Thien-Thi Nguyen wrote: > () Patrice Dumas <[email protected]> > () Tue, 4 Dec 2012 22:30:51 +0100 > > To be able to format a printindex rapidly, the index entries should > be easy to access directly, so, in my opinion all should be in the > file (duplicated) with the beginnning and end offsets just like > nodes. > > IXIN 1.3 is available: <http://www.gnuvola.org/software/ixin/>. It has > section tree and two-letter indices (NB: non-duplicated), but not tweaks Indeed, if the node is in main text, no need to duplicate. Yet, if I have read correctly, in the *** tli (zero or more) you duplicate the index entries in TERM, although it could also be an offset and a length. Not really a big deal. > - META-divvying work-in-progress > > See new header "groupings" in README. Looks good. codequotebacktick codequoteundirected txideftypefnnl have a non set form, and it should be that way, using @set in place of commands is not what we do today. Regarding your interrogation about @documentencoding, @documentencoding indeed should in general be unique. However, texi2any should behave correctly if another part of the document has a different documentencoding. I guess that TeX may be less forgiving. In any case, I think that IXIN should only try to handle files utf8 encoded, whatever the documentencoding (or use the *** coding you propose) and so should not really do anything upon encountering. documentlanguage should certainly also be in the XID, to have a language to start with, as it may be needed even before the meta section, for instance for title, title page, copying.... > I wonder if there is a better term than "section tree". The manual uses "chapter structuring". > Another, more > important, doubt is how to make sure that functions in the ‘cp’ index > (say) are formatted with ‘@tt’ as opposed to ‘@r’? That information > doesn't seem to be captured anywhere. That's indeed an interesting information that should be added, especially since it is not necessarily easy to find out because of merged indices entries. Also the index the entry is finally associated with, if there are indices merged with @syn*index. You still miss something similar with TLI handling for listoffloats / floats. Regarding macro, I think you don't need them. They should be fully expanded (as well as @value) in the IXIN document. Last, about xml to sxml using the perl module, I'll have a look. -- Pat
