On Sat, 29 Sep 2001, Keith Packard wrote: > > I think finally implementing the subset mechanism for BDF/PCF would > > provide the cheapest and most effective improvement that can be done on > > the XFontStruct front for ISO10646-1 fonts. > > It's hard to imagine how subsetting will help; few applications know the > complete range of glyphs they'll use ahead of time. Of course, they could > examine the locale and guess, but that's expecting a lot from the > developers when the easy solution is just to load the whole font.
I still have a bit the suspicion that font merging is primarily needed to satisfactorily display pathologic pages such as http://www.columbia.edu/kermit/utf8.html which are certainly cool demos, but they mix far too many scripts to be a realistic and in any way representative application demonstration. What fraction of X11 users exchange email with the editors of Japanese<->Ethiopian dictionary editors on a regular basis? Usually, users are prefectly able to pick the fonts manually that have the coverage and style they really need, and if a glyph is missing from that font, it will typically also not be readable for the user. Mixing fonts automatically can lead often to far more ugly results than had the application just stuck with a single font and used its default glyph for everything missing. The urge for implementing the BDF subsetting is the following: X.Org wants to include my -misc-fixed-*-iso10646-1 fonts into their next release, which would be a major step towards having UTF-8 xterm spread throughout the commercial Unix world as well. But they are currently somewhat confused about my bdftruncate.pl hack, with which the fonts that we install are cut to 0x0000-0x31ff in order to keep XFontStruct from exploding completely. If we had the BDF subsetting implemented, we could immediately stop using bdftruncate,pl, because now applications can decide trivially at font load time whether to truncate the font ("*-iso10646-1[0x0000_0x31ff]") or not ("*-iso10646-1"), and then the coverage provided by the core fonts on a server would a bit less. Terminal emulator users are not interested in glyph merging, because they typically use bitmap fonts that contain already all glyphs available and are sensible for a given font size. But terminal emulator users are concerned about the startup time of xterm, which increases when a huge XFontStruct has to come over the network cable, especially on the zillions of X terminals out there who don't have and won't have RENDER available, because nobody updates the X server ROM, but that use remote font servers that still can be updated. Many university libraries with old X11 online catalog interfaces are examples that could benefit greately from minor ISO10646-1 related improvements in the XLFD font server world. Markus -- Markus G. Kuhn, Computer Laboratory, University of Cambridge, UK Email: mkuhn at acm.org, WWW: <http://www.cl.cam.ac.uk/~mgk25/> _______________________________________________ I18n mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/i18n
