Subject: Re: PATCH: Redhat 7.0 compile fixes for very latest CJK.
From: Vlad Harchev (hvv@hippo.ru)
Date: Tue Oct 31 2000 - 03:15:40 CST
On Tue, 31 Oct 2000, Martin Sevior wrote:
>
> On Tue, 31 Oct 2000, Vlad Harchev wrote:
>
> > On Tue, 31 Oct 2000, Martin Sevior wrote:
> > >
> > > On Tue, 31 Oct 2000, Vlad Harchev wrote:
> > >
> > > >
> > > > No, 'char' is treated as 'signed char' per ANSI C standard. So (unsigned
> > > > char) cast was necessary.
> > >
> > > Thanks for the clue. The high order bit was propagated into the second
> > > byte.
> > >
> > > >
> > > > > I've applied the patch to a new tree merged with Dom's new gnome-print.
> > > > > It compiles and builds without problems on RH 6.2 in unix. I haven't
> > > > > tried a gnome build yet coz I gotta get the latest gnome-print. Now the
> > > > > Lists dialog and the symbols inserted into frame look correct but printing
> > > > > the symbols doesn't work. I get some strange symbol instead. When I used
> > > > > "Insert Symbol" to insert "Dingbats" fonts into the text they appeared OK
> > > > > on the screen but upon printing looking like Chinese characters.
> > > > >
> > > > > The old abi did this too.
> > > >
> > > > OK, I'll look into it and it seems it would be easy to fix it.
> > > > measureUnremappedChar is guilty.
> > > > But did this problem exist between my non-latin1 support patch?
> > > >
> > >
> > > Yes it did. I should have told you sooner but you seemed to have enough
> > > other things to do.
> >
> > I assume you did s/between/before/ when answering my question? If not,
> > please tell whether this problem was before (not "between") my first
> > non-latin1 support patch.
> >
>
> I mean your original patch caused this problem. I had not noticed it
> before your patch for Russian support.
Hmm. I've looked at the sources and have a *very very strong* impression that
it wouldn't work with any version of AW (and before my patch) with dingbats
font. The reason it doesn't work is that encoding of that font is
'adobe-fontspecific' and 'sym_enc' table is used in
XAP_UnixFont::getMetricsData for mapping unicode values to indices. But since
there are no glyphs with names as in table 'sym_enc', most characters (except
space :) are considered non-existant.
So, I will add the logic that will count how many glyphs under standard names
are found in font, and if the percentage is < 80% of glyphs in font, the
indexes will be used instead of glyph names - i.e. all will work fine with
such font.
I should say that printing chars in dingbats fonts works fine when done under
russian locale since no "map by name" is engaged under it.
I will prepare a patch for that.
> Martin
>
Best regards,
-Vlad
This archive was generated by hypermail 2b25 : Tue Oct 31 2000 - 03:33:58 CST