Jukka said: > The comments at the start of NamesList.txt say that it is > “semi-automatically derived from UnicodeData.txt”, but the information > you are referring to has actually been picked up from the code charts. > They contain both informative alias names and cross references.
The "semi-automatically derived" part refers to the way NamesList.txt is generated. It is actually the algorithmic output of a merge tool that takes UnicodeData.txt, StandardizedVariants.txt, and a master annotation markup file as input. The automatic part is the algorithmic merge and the reliance on UnicodeData.txt itself, so there will be no errors in the normative part of the NamesList.txt content, i.e., most importantly the character names, their code points, and their decomposition mappings. The non-automatic part is the necessity of maintaining the master annotation markup file manually. As new aliases, cross-references, and other annotations are proposed, they are reviewed by the editorial committee, and if they seem o.k., are added to the master annotation markup file. The code chart production is then driven by using NamesList.txt as input, along with another set of manually maintained input files which define font choice and glyph mappings to the Unicode code points. The tool that does that part of the production is our friend "unibook". > > Since HYPHEN BULLET and HYPHEN can hardly be considered as aliases of > each other, the issue is whether a cross-reference is useful. Leif's wording was a little garbled, but I agree it comes down to the issue of adding a cross-reference between the two. It seems harmless and perhaps useful, so I suspect that the editorial committee will end up adding it. --Ken
