Re: [Devel] Postscript hinter module commited

[Vadim Plessky]

On Tuesday 09 October 2001 07:03, David Turner wrote:
|   Hi Vadim,

Hi David!

|   Vadim Plessky a écrit :
|   > (development version) .  So all people testing cooker/rawhide can
|   > (potentially) send bug reports.
|
|   The plan is the following:
|
|    - make a 2.0.5 release quickly. This one should only contain
|      bug fixes for 2.0.4 (actually, many of them :o)
|
|    - merge the two branches, and solve any problems that might
|      happen. I suspect many will, even with Emacs "help"

Does this mean that after 2.0.5 release "FT with the Postscript hinter" 
becomes latest development branch?
In this case, I vote for *soonest* release of 2.0.5 :-)
Will it be possible to run 2.0.5 and this development branch simultaneously?
|
|    - release 2.0.6 with the Postscript hinter, and probably a
|      few improvements to the auto-hinter. Possibly some important
|      bug fixes too.. in this version, the "type1", "cid" and "cff"
|      font drivers will all use the Postscript hinter..

This is amaazing!..

|    - then, make a few internal changes (mostly changing the internal
|      interface and structure of modules, in order to simplify a small
|      number of important operations, as well as solve difficult bugs
|      like "no embedded bitmap is loaded when the auto-hinter is used")
|
|      I would also like to write a new monochrome rasterizer for
|      Postscript formats, mainly because hinting diagonals correctly
|      is simply too complex and costly to be done in real time..

I think auto-hinting in real-time is just not possible.
It doesn't depend on case you have just vertical/horiz. stems or diagonal as 
well.
But, from other side, number of glyphs with diagonal elements in 
English/Latin alphabet is rather limited. So it still can be good solution.
As for Cyrillic, number of glyphs requiring diagonal hinting is much higher, 
and it can really become performance penalty.

Do you auto-hint a font as a whole (probably, answer is yes, otherwise there 
is no way to calculate array of standard/typical H and W stem widths)?
In this case, I think, you still need some mechanism to turn off auto-hinting 
for some Unicode ranges.
For example, standard Arila or TimesNewRoman from Win98 have 1296 glyphs 
each, while most likely typical user uses just 128. And I doubt you can 
auto-hint Arabic or Hebrew glyphs. :-((

There were several reports on Koffice/kde-devel lists that with some fonts 
KWord was becoming non-responsive. In turn, it was found that this guy used 
23MB Arial TTF, with around 40 000 glyphs (from Office XP Proofing Tools, or 
something like that)! And response time, indeed, was 7-8 sec for line 
update... It can be somewhat XFree86 or QT problem (rendering of all glyphs 
at once, etc.), but even auto-hinting of 40000-glyphs will take a lot of 
time...

|
|    - make a 2.1 release..
|
|   Regards,
|
|   - David
|
|   _______________________________________________
|   Devel mailing list
|   address@hidden
|   http://www.freetype.org/mailman/listinfo/devel

-- 

Vadim Plessky
http://kde2.newmail.ru  (English)
33 Window Decorations and 6 Widget Styles for KDE
http://kde2.newmail.ru/kde_themes.html
KDE mini-Themes
http://kde2.newmail.ru/themes/