Re: [Groff] mapping of glyphs to Unicode

[James Cloos]

>>>>> "Bruno" == Bruno Haible <address@hidden> writes:

Bruno> Can you explain more? Symbol fonts have only one 'angleleft' and only one
Bruno> 'angleright' glyph. Therefore for the 'devps' device it is irrelevant
Bruno> which Unicode code points we use - \(la must map to 'angleleft' and
Bruno> \(ra must map to 'angleright'. So for the 'devps' device the Unicode
Bruno> code points are groff internal.

My point was that given that la and ra point to glyphs from the Symbol
font in devps, that devutf8’s author probably presumed they should
then map to the iso10646 code points which were added for
compatability with the Synbol font, rather than the similar text
characters.  One can see how one may presume that anything coming from
Symbol was intended for a math context....

Bruno> IMO what we do in the devutf8 and devhtml devices should not be
Bruno> influenced by Adobe glyph list considerations.

Agreed.  I was describing my presumtion of the author's thinking, and
of the unicode/10646 thinking.

Part of the problem of course is that unicode/10646 deals with
characters and the dev filesets were at least originally designed to
deal with glyphs.  That devutf8 and devhtml should map to characters
rather than glyphs is true.  How easy that will be in the general
case — where unification vs. disunification, nfc vs nfd, et al all
come into play — makes for Interesting Times.

-JimC
-- 
James H. Cloos, Jr. <address@hidden>