From: Asmus Freytag (email@example.com)
Date: Sun Nov 25 2007 - 16:38:11 CST
On 11/24/2007 3:50 PM, Elliotte Harold wrote:
> James Kass wrote:
>> People generate PDFs from plain text and also from HTML. If I were
>> using HTML for the purpose of reproducing a page from a two hundred
>> year old book, I would want to preserve the ligature information. As
>> something of a purist, I would want my reproduction to have ligatures
>> where the original had ligatures, and not have ligatures where the
>> original hadn't any.
> If you require that level of typographic fidelity, you wouldn't be
> using HTML. Ligatures would be the least of your worries in that
If that book was a 200-year old German book, you may very well want to
capture it in the Fraktur style it was published in. In that case, you
can't get around ligatures (both mandatory and prohibited), even if your
intent is not to capture the *actual* ligatures used by the particular
typesetter -- it's simply that the style as such requires some ligatures
and usually uses a large number of others. I believe that some of the
latter could be (and were) used, or not used, to allow better line
justification (but I don't have ready evidence available for that one
way or the other).
Books from that period would alternate between Fraktur (bulk of the
text) and Antiqua/Roman style (lain or foreign words) in a way that in
some way reflected the choices of the author (just like details of
spelling or punctuation). Rendering such works in modern style
essentially discards such information.
Therefore, even if you are only interested in getting at the *text* so
to speak, you end up having to have support for ligatures (and control
over them, because of the twin issues of required and prohibited
ligatures). That level support is in principle possible with standard
rendering layers, given that you have an OpenType font with the correct
ligature tables. and that your text uses ZWNJ for the prohibited
ligatures. So that's all very much plain text and HTML.
If you want to go further and actually create a faithful record of the
*typography* of such texts, your requirements would go beyond what you
can do with OpenType and ZWNJ, you would need actual markup.
This archive was generated by hypermail 2.1.5 : Sun Nov 25 2007 - 16:41:16 CST