i'm facing a problem with the seeming lack of any way to encode correct glyph-combining information in character-cell bdf bitmap fonts for use in x. i've heard of people doing horrible hacks like just making the combining characters zero-width and offset to the left of the cell, so that they naturally overstrike, but this has several major problems that cannot be solved:
- it only works for programs that render a whole string with x, not actual character-cell based programs that will draw and redraw characters in their respective cells, i.e. terminal emulators. - it does not allow vertical stacking of combining marks, i.e. if there's more than one they'll just overstrike one another and become illegible. - it does not account for the fact that the base character may have to change form to accomodate the combining characters (either due to conventions of the script itself or limitations of the font size, e.g. uppercase latin characters having to become shorter to make space for accents in small fonts). what i would like is a standardized system for (ab)using or extending the x11 font metrics to store the data on how characters should combine. unfortunately i don't yet understand the x font system well enough; however i believe that the width/ascent/descent/etc information, which are otherwise not very useful on combining characters, could be used to store the offsets for rendering the combined character in the correct position relative to the base character or the previous combining character. this rendering could be performed one-character-at-a-time by the application, and possibly also by an extension to the x server. as for alternate glyphs for the base character when it's used in combinations, i don't know what to do since x seems to have a shortage of glyph numbers already. the only alternative i know is for unicode terminals to ignore the whole x font system and load their own pixmaps and data tables of combining information. i don't think anyone particularly likes this solution, but perhaps i'm mistaken. other ideas...? rich -- Linux-UTF8: i18n of Linux on all levels Archive: http://mail.nl.linux.org/linux-utf8/
