[Devel] RE: [Freetype] Pango or Freetype and some other questions!!

[786]

Hi, 

I came across a small problem in working of the Chaining Contextual
Substitution.

A little overview into the specification; during text processing, a lookup
is applied to each glyph in the string before moving to the next lookup. A
lookup say L1 is finished for a glyph after the client locates the target
glyph satisfying the backtrack and/or lookahead context and performs a
substitution and moves onto the next glyph in the sequence. Now the problem;
Well in the way how it is done in freetype, the substitution output say X is
actually copied on to the ANOTHER output array (out).  It is only after the
same lookup is tried on the remaining input (in) string that the
substitution output X is copied back on to the in string. 

As a result, during the time the lookup L1 is being applied on the remaining
input string, it is in a way 'blind' to the substitution that it (itself)
had earlier done. And in this way the chaining part of the 'chaining context
subst table' is choked.

So here is the solution for ccsf3; same can be done for ccsf1 and ccsf2.
In static TT_Error  Lookup_ChainContextSubst3  Instead of....

if ( bgc > in->pos || in->pos + igc + lgc > in->length )
        return TTO_Err_Not_Covered;
      if ( bgc )
      {
         curr_pos = 0;
         s_in     = &in->string[curr_pos];
         bc       = ccsf3->BacktrackCoverage;

      for ( i = bgc, j = in->pos - 1; i > 0; i--, j-- )
      {
        while ( CHECK_Property( gdef, s_in[j], flags, &property ) )
        {
           AND SO ON....

Only for the backtrack context, the out array should be used. That is...

             HERE
if ( bgc > out->pos || in->pos + igc + lgc > in->length )
        return TTO_Err_Not_Covered;
      if ( bgc )
      {
          curr_pos = 0;
          /*Earlier */ 
          /* s_in     = &in->string[curr_pos]; */
          /*Now */
          s_in     = &out->string[curr_pos];
          bc       = ccsf3->BacktrackCoverage;

                              HERE
      for ( i = bgc, j = out->pos - 1; i > 0; i--, j-- )
      {
        while ( CHECK_Property( gdef, s_in[j], flags, &property ) )
        {
           AND SO ON....

Aamir



-----Original Message-----
From: 786 
Sent: Thursday, June 03, 2004 3:05 PM
To: 'Owen Taylor'; 'address@hidden'
Subject: [Freetype] Pango or Freetype and some other questions!!


/duplicate. Note in previous it was not 'offline' but *offlist*

Hello,

Long long time ago there was this conversation....(read below if you have
time).

Well I had a small question in mind. But first consider this Mr Lemberg said
that "not aware of any parts of the substitution specification that the
PANGO CODE (or that matter the opentype code), doesn't implement" (Note the
capital PANGO)

" What about the new GSUB lookup type 8?  AFAIK, this is specifically
for Urdu.  Have you added it to to PANGO?" (Note the capital PANGO)

Well what I don't understand is why are we calling it Pango code? Shouldn't
it be freetype's code (ftxgsub.c in freetype package)? Pango uses this code
after some initial 'script specific processing' when it actually applies the
rules on 'tagged' input string. So this does not make it pango code. As a
result when we talk about adding GSUB type 8 don't we actually mean in
ftxgsub.c.

>> Have you added it to to PANGO?" 
So shouldn't it be the freetype guys who should be thinking about adding it?
And if it does have to be added then it's right in the OT code in pango
where other types are implemented. 

Another question: Say for the moment 'All the GSUB tables has been
implemented' is there a any list of which features have so far been
implemented. Is calt implemented? I have pango-1.2.5 & Freetype-current,
this does not?


And finally Pango-1.2.5 does not implement calt, that's no problem..it can
easily be added and tested in gedit. Calt uses the LookupType 6: Chaining
Contextual Substitution in ftxgsub.c. But there seem to be some problem; the
calt rules listed in the font do not seem to apply at some instances. Now
this could either be the problem with my gedit/linux or with the lookuptype
6's implementation. Can Freetype's team allow me to please discuss this
problem with them offlIST?

Thanks

Aamir


-----Original Message-----
From: Owen Taylor [mailto:address@hidden 
Sent: Tuesday, October 07, 2003 1:57 AM
To: address@hidden
Cc: address@hidden
Subject: Re: [Freetype] Add more claws to FreeType.

On Sun, 2003-10-05 at 04:54, Werner LEMBERG wrote:
> > I'm not aware of any parts of the substitution specification that
> > the Pango code (or that matter the opentype code), doesn't
> > implement.  For positioning, cursive attachments of vertical text is
> > not implemented, but even for Mongolian that probably isn't big
> > deal.
> 
> What about the new GSUB lookup type 8?  AFAIK, this is specifically
> for Urdu.  Have you added it to to Pango?

No, I wasn't aware of that.. it didn't come up in any fonts I've
tested widh. It doesn't look that complicated to implement, if
someone was interested.

Regards,
                                        Owen


_______________________________________________
Freetype mailing list
address@hidden
http://www.freetype.org/mailman/listinfo/freetype