Initially I just wanted keyword substitution inside a .xltm file. But
really, Office 2007 documents are just zip files, and that got me
thinking that the problem is more general.
Many of the best features of subversion only operate on text.
Unfortunately, there are tons of compound document formats out there
that actually *are* text, when you get right down to it. The text just
happens to be compressed and collected with some other stuff.
So... what if a compressed archive was extracted during checkin, and
re-archived on checkout? This would
(a) more easily allow for text diffs between versions of documents.
(b) permit keyword substitution
(c) allow the choice of other, more effective compression (eg, bz2) on
the back end.
Thoughts?
