[XeTeX] Problem with ocrb10.otf ligature 'fi'

[Pander]

TeX Live list members: see full thread here:
http://tug.org/pipermail/xetex/2011-June/020681.html for now keep the
discussion at XeTeX's list.

On 2011-06-13 14:22, mskala at ansuz.sooke.bc.ca wrote:
> On Mon, 13 Jun 2011, Pander wrote:
>> TeX Live 2010
>>
>> /usr/local/texlive/2010/texmf-dist/fonts/opentype/public/ocr-b-outline/ocrb10.otf
> 
> That is Zdeněk Wagner's auto-conversion of Norbert Schwarz's Metafont
> source.  It doesn't contain f-ligatures no matter what the GSUB table may
> say.  I took a look at it with Fontforge and I see that it contains a GSUB
> table pointing the ligatures at "alternate" and added non-ASCII characters
> from the Schwarz version, some of which happen to be ligature-like but not
> the correct ones.  For instance, "fl" points at the Æ glyph.
> 
> I recogize that pattern because it happened in an earlier version of my
> own version of the font, as a result of auto-conversion.  The thing is,
> Schwarz's Metafont files used a nonstandard custom encoding.  If you
> simply convert the font code point for code point to whatever the default
> 8-bit Adobe encoding might be, you end up with Schwarz's extra glyphs at
> the "f-ligature" code points (as well as some distortions at quotation
> mark, dotless i and j, and similar code points).  The existence of a GSUB
> table pointing at those points can probably be explained by defaults from
> the auto-conversion.  So in summary, yes, it's a bug in the font.

Could the conversion software generate a warning when it recognises such
a situation?

> The current version of my own OCR B fonts, available on ansuz.sooke.bc.ca,

http://ansuz.sooke.bc.ca/page/fonts

> is also based on Schwarz's, but via a more manual conversion process
> (rewriting the Metafont sources to work with MetaType1), and I've
> attempted to put all glyphs at their correct Unicode code points.  It
> contains a GSUB table for alternate forms of glyphs, but none for
> ligatures.
> 
>>> I just downloaded the demo from here:
>>>       http://www.barcodesoft.com/ocr_font.aspx
> 
>> Maybe TeX Live should use these OTF files?
> 
> Barcodesoft's "free" version is a watermarked demo of an expensive
> commercial product, basically just an advertisement, and for that reason I
> wouldn't recommend its distribution in TeXLive; I'm not even sure that the
> license agreement would allow such distribution.

In effect it is freeware and is owned by Barcodesoft. But according to
your README, one is allowed to redistribute this and your enhanced
version. So in the same way would TeX Live be able to so. The metadata
in the font files provides proper credits.

I think, first CTAN needs to be properly updated, see:
  http://ctan.org/search/?search=ocr&search_type=description
Probably many of these CTAN package can merge.

Subsequently TeX Live can do their update. For now, I'll forward this
also to them.

Would it also be possible to generate Bold, Italic, Light and Condensed
versions for OCR-A and OCR-B? In that way it is also backwards
compatible with the current OCRA fonts.

> 
> 
> 
> 
> 
> --------------------------------------------------
> Subscriptions, Archive, and List information, etc.:
>   http://tug.org/mailman/listinfo/xetex