At 12:48 +0330 2001/02/13, Roozbeh Pournader wrote: >In Persian, we usually do not have the three classic families. In Iran, >there is rarely a need for typewriter style, that's only used for Latin >texts, and in the case they really need to show Persian on the screen, >they use screenshots. Also, I know only three non-bitmap typewriter fonts, >and the only almost-free one is MS Courier. I use one of the other two >when writing a manual, but I'm among the few who use such a thing. I am not sure that the NFSS model is so suitable for European languages & Latin alphabets either, as it is quite difficult to find a good set of fonts matching each other except for the NFSS attributes. In math mode, it is of course important to have a set matching each other in wehat can be thought of appear side-by-side in math formulas. Instead (as an input), I tend to think of the fonts as having a set of attributes. If one group together some fonts together with different attributes, one might get a font-family. But different font families may have different sets of attributes available. Here are some attributes that come to my mind: ligatures: sans-serif, serif weight: thin, medium, bold, heavy slantedness: left, upright, right monospace: false, true extendedness: narrow, condensed, normal, extended outline: false, true shaded: false, true A full attribute would be a choice of tuples (ligatures, weight, slantedness, monospace, extendedness, outline, shaded) and a set of fonts with different atributes would be a font family. Some attributes may be needed to be made more detailed by more choices. >This the Persian model only. Then you need Latin. For each of the styles >that also may contain Latin text, they should use an equivalent fonts >that goes with it. There are some difficulties here also, one of them >being the different direction and hence different slants: there are few >backslanted Latin fonts, so you need to use slanted or italic Latin with >backslanted Persian (which is as ugly as any backslanted font in any >script). So this might be possible then with such a model. Hans Aberg