Adam, I feel like a complete idiot now. I had been so proud about this idea, but after re-reading you MyWay about OpenType, I see that I had been reinventing the wheel: this is exactly the solution you had been suggesting almost two years ago. Thanks for being generous about this... However, your post made me think: I know nothing about XSLT, but enough perl to shoot myself in the foot. I guess if I had a version of texnansi.enc with the unicode values in addition to the names, that would be a good starting point. I was thinking of this route: 1. use ftxdumperfuser to produce cmap.xml, 2. use perl to reduce it to two values: glyphName % UNICODE_VALUE 3. use perl to extract the lines corresponding to a given encoding and put them in the right order. Sounds feasible? Do you know where I could get such a unicode-aware version of texnansi.enc? Best Thomas On Feb 27, 2005, at 10:39 AM, Adam Lindsay wrote:
Indeed, that's one of the reasons why I came up with the unicode ("symbol"[1]) scripts... there are common utilities (ttx and Apple's ftx suite) that work well at associating canonical characters with glyph names specific to a font.
I'm sure some enterprising XSLT hacker could take my scripts as a starting point and make them work with specific TeXy encodings, not just individual Unicode vectors.