[links-list] Re: Links 2.0pre6 - Cyrillic :-(

aludal aludal at softhome.net
Mon Nov 4 00:19:59 PST 2002

В сообщении от 3 Ноябрь 2002 23:04 вы написали:
> > В сообщении от 3 Ноябрь 2002 17:16 вы написали:
> > > No El, no U, no De, no Zhe, no short I in links -g - neither small
> > > nor capital ones!
> >
> > I have a complete Russian Cyrillic set (working on full Cyrillic set
> > right now, and on "bolds".... wish they'd have italics on Links...)
> Great! I'm looking forward to it.
So, it makes two of us. Not more yet, as I didn't hear a word from Czech Links 

> > > At the moment, links -g has better support for Hiragana/Katakana than
> > > for Cyrillic :-(
> >
> > Once again: who needs that Hiragana/Katakana without kanji, tell me,
> > please. Moreover, I've got relatively full (CJK included) Bitstream
> > Cyberbit on this very Linux box I use Links, who's gonna cut it,
> > re-outline, hint, and antialiase more than 35 thousand xxxxx.png files?
> Antialias is not necessary for Kanji, I guess. I'd be content with Japanese
> pages I could at least *read* instead of seeing trash. So maybe these pngs
> could be generated automatically to render simply a *readable* version of
> such pages?
"Antialias" is too big a word for simply making a CJK's glyph readable in 
16-18, or even 20 pointsize bitmap of png. Nevertheless, even if it's reduced 
to more or less practical hinting (in terms of legibility), it's a formidable 
task. There's a scant possibility to automate this process, as was mentioned 
to me by George Williams (author of PfaEdit) in correspondence. His last 
version of excellent type editor is capable of executing png-producing 
scripts over a given (Unicode) font. The major problem here is that scripting 
language is quite poor, and the resulting png images:
a) are plain squares of, say, 100x100 pixels, with outlines in them deformed 
to touch all 4 sides of such a square: squeezing, or shaping them back into 
more or less initial form of a glyph cell affects hinting very much;
b) have no baseline information embedded, or even returned as a separate 
parameter for post-editing. A bdf efont may have baseline data retained in 
similar conversion, but it was just George Williams' initial guess;
c) come out non-inverted, so a usual non-symmetrical antialiasing mishapes 
outlines badly.
All of this means a LOT of manual post-production retouching for every single 
png out of 37 or 40 thousand of them. But OK, if you're so much interested, 
why don't you grab the latest version of PfaEdit, then Titus Cyberbit Basic, 
or Bitstream Cyberbit, or even efont bdf (this one is looking very nice for a 
discerning eye), and do the scripting? PfaEdit is well documented on these 
scripts. Besides, you could make a research on post-production scripting in, 
say, Gimp, to lessen the burden of manual retouch.
But don't ask me to do it: I spent like 3 complete weeks in redesigning and 
retouching a mere 66-letter set of Russian Cyrillics, and it's hard for me to 
believe that I need to go through ~200 more -- let alone CJK, or whatever.
Besides, I'm of an opinion that the whole task of "embedding" 12, or 15, or 40 
MByte (in case of Arial Unicode) of font into tiny Links is crazy. Even 3.5 
Mbyte of Titus (no CJK) is crazy enough.

> > So, when and if Links
> > authors ever try to make Links fully Unicode/Web encodings compatible (it
> > means, among others, a *correct* support of at least THREE Japanese
> > codepages), that Links version (my guess it would sound something like
> > 3.1pre12, hehe....) would be more or less freetype2-compatible for
> > rasterizing trutypes and capable of using mozilla layout engine.
> > "Brain" guy would give me his permission, and I'd  upload my re-designed
> > and re-hinted Russian pngs. As soon as, that is... Stimmt?
> >
> > > "Mahanika by L.D.Landau i E.M. Lifshic" becomes "Mehanika by . . an a
> > > i E.M. i ic". ARGL :-(
> >
> > So, what's your problem with 'Mehanika'? Out of these two, it's exactly
> > 'Mahanica' which is wrong transliteration.
> Apart from that sloppy use of "transliteration" -- I guess that was
> simply a slip of the finger.
whose finger slipped, I wonder? hehe.... BTW, "Kurs Teoreticheskoy Fiziki" by 
Lev Davidovich Landau and Evgeny Mihailovich Lifshits,  of which "Mehanika" 
was the 1st volume, was my standard textbook in my alma mater 30 years ago.

--- Arbeit macht frei Software ---

Unsubscribe: send email to links-list-request at linuxfromscratch.org
and put unsubscribe in the subject header of the message

More information about the links-list mailing list