[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: hidden composites

Thierry Bouche <Thierry.Bouche@ujf-grenoble.fr> writes:
> Hi again & thanks!
> nice explaination, I hadn't recognise these computations!
> maybe it's doable in postscript?

Two difficulties for doing the computations in PostScript are that read
access to the CharStrings dictionary of a font is usually forbidden,
and one would have to decode the bytecodes that make up the CharSrings
entries. There may be ways around this, but it is likely more trouble
than its worth if you have t1disasm and perl handy.

Hilmar Schlegel <hshlgaii@mailszrz.zrz.TU-Berlin.DE> writes:
> Good suggestion - the source [for the partial font downloader in
> pdftex] is avail for your attempts in this direction ;-)

I've already fixed bugs in the dvips partial font downloader, as some
here know, so it's not so outlandish for me to suggest that this bug
might be fixable. Currently, I don't use pdftex nor the troublesome
fonts in question, so I have no burning personal need to fix the bug.

> Even more specifically, the tools for generating the complete (and
> exact) AFM are available for years meanwhile. But of course please feel
> free to reinvent the wheel...

I'm not aware of any tools that generate the same `CC' entries and the
same kern pairs provided by the font vendor (which is the only ``correct''
AFM for a font). It is my understanding that kern pairs are usually hand
tuned by the font designer, so I do not see how what you say can be

Nevertheless, if you feel I am reinventing the wheel, do please tell us
all of the utilities that are available that Thierry should be using to
solve his problem.

I wrote:
>> Anyway, I hope this helps,

... and Hilmar Schlegel replied:
> I fear not too much - as long as getafm or similar code is not adjusted
> nothing will change.

Thierry seemed very pleased, and he was the person I was trying to help.

Getafm has nothing to do with this, as far as I can see. Thierry had
the vendor's AFM file, which is more complete than anything getafm might
produce -- especially since getafm lists no kern pairs and no composite
character entries.


P.S. Having corresponded with Hilmar Schlegel once before and been hurt
by what I found an arrogant and condescending attitude, and disappointed
at what I saw as lack comprehension and insight, I will likely not
respond to anything else he might say on this matter.