Re: blurry fonts (again)

[Celejar <celejar@gmail.com>]

On Fri, 6 Nov 2009 00:03:39 +0100
Florian Kulzer <florian.kulzer+debian@icfo.es> wrote:

...

> OK, so let's recap:
> 
> You print the Hebrew Wikipedia homepage with "print to file" in
> iceweasel. You get a PDF with misaligned Hebrew characters (kerning
> seems to be off) and this font information:
> 
> name                                 type              emb sub uni object ID
> ------------------------------------ ----------------- --- --- --- ---------
> FrankRuehlCLM                        Type 1            yes no  yes      5  0
> FrankRuehlCLM                        Type 1            yes no  yes      7  0
> BitstreamVeraSerif                   CID TrueType      yes no  yes     37  0
> 
> I do the same thing and I get a PDF with properly aligned Hebrew
> characters and this font information:
> 
> name                                 type              emb sub uni object ID
> ------------------------------------ ----------------- --- --- --- ---------
> TimesNewRomanBold                    CID TrueType      yes no  yes      5  0
> TimesNewRoman                        CID TrueType      yes no  yes      7  0
> DejaVuSans                           CID TrueType      yes no  yes     38  0
> 
> I checked my "good" PDF with pdfedit and the Hebrew characters are
> indeed typeset in TimesNewRoman and TimesNewRomanBold. DejaVuSans is
> used only for the extra information in headers and footers (URL, page
> number, date, time). Your "bad" PDF uses the FrankRuehlCLM font for the
> Hebrew text and BitstreamVeraSerif for headers and footers.
> 
> I think there could be a problem with the FrankRuehlCLM font (package
> "culmus") or pango has a bug with Type 1 fonts. You could try to
> downgrade libpango1.0-0 and friends to their Lenny versions (provided
> that this does not break anything else on your system, of course) or you
> could remove the culmus package so that icewesel is forced to use e.g.
> the DejaVuSerif font for Hebrew passages. (The DejaVu fonts are extended
> versions of their BitstreamVera ancestors; they include Hebrew
> charcters.)

Okay.  So I tried removing culmus, and now a reasonably legible and
normal looking, if not particularly attractive (although that may be
purely subjective, and in the eye of the beholder) PDF is generated,
for which pdffonts shows:

name                                 type              emb sub uni object ID
------------------------------------ ----------------- --- --- --- ---------
FreeSerifBold                        CID TrueType      yes no  yes      5  0
FreeSerif                            CID TrueType      yes no  yes      7  0
BitstreamVeraSerif                   CID TrueType      yes no  yes     35  0

I tried using fc-match on these font names, as you showed me in our
first iteration of this thread, but I'm a bit confused by the output:

~$ fc-match FreeSerif FreeSerif.ttf: "FreeSerif" "Mittel"

This seems to indicate that FreeSerif comes from the FreeSerif.ttf
file, which is from the ttf-freefont package. OTOH:

~$ fc-match FreeSerifBold Vera.ttf: "Bitstream Vera Sans" "Roman"

Does this mean that FreeSerifBold is in Vera.ttf and not in
ttf-freefont?  But:

~$ dpkg -L ttf-freefont | grep FreeSerif
/usr/share/fonts/truetype/freefont/FreeSerifItalic.ttf
/usr/share/fonts/truetype/freefont/FreeSerifBoldItalic.ttf
/usr/share/fonts/truetype/freefont/FreeSerif.ttf
/usr/share/fonts/truetype/freefont/FreeSerifBold.ttf

I find all this stuff fairly baffling.

Regarding your other suggestion about downgrading libpango, trying to
do that requires removing some packages that are important to me, or
doing a fairly significant looking downgrade of a number of packages,
so I'm going to hold off on that, at least for now.

Thanks much!

Celejar
-- 
foffl.sourceforge.net - Feeds OFFLine, an offline RSS/Atom aggregator
mailmin.sourceforge.net - remote access via secure (OpenPGP) email
ssuds.sourceforge.net - A Simple Sudoku Solver and Generator