From: Kent Karlsson (kent.karlsson14@comhem.se)
Date: Tue Jan 10 2006 - 16:25:55 CST
[Not citing all of the message, but my comment below applies generally;
and
I've said it many times before. Unfortunately it seems to need saying
yet again.]
> > Likewise for V and U. Or would you encode the V in that stamp sample
as
> > U? I do hope not.
>
> If I wanted the text to be searchable, spell-checkable and
> sortable as standard German text, you bet I would.
I think it is the AUTHOR of the text that should decide on all spelling
issues.
That includes diaeresis vs. e above, U vs. V and many many other cases
(including
numerous spelling reforms for numerous languages). The font designer
should
have NO say in matters of spelling, i.e. the font should not give the
*appearance*
of a spelling different from that actually used in the text. E.g. IMO it
is not acceptable
to actually have the character sequence for "vila" (modern spelling) be
displayed
as "whila" (old spelling), or display "de har" (modern spelling) as "de
hava" (old
spelling, for an old pronunciation), or display "Ni" (modern spelling)
as "I" (old
spelling, again for an old pronunciation), etc., etc.
So, yes, if you want to apply a programmatic spell checker, it has to be
for the
orthography used; not trying to apply a spell checker for modern
spelling on
what looks like old spelling (but just looks that way via a
really-bad-idea font
applied to modern spelling).
That it is the case that some (acceptable) font variants have separate
encodings is
not a licence to do apparent spelling substitutions via the font, even
for simple cases.
"Artsy" text display can do whatever, and use whatever encoding ("hacked
fonts"),
or even be available solely as images (i.e. not as character strings
from which the
images are made via a font; though there may be an "alt" text, like for
HTML images).
I would not count the example cited a necessarily "artsy", though given
as an
image only. The (apparent; I guess the stamp may be hand drawn) font for
the
text is not very "artsy", I think.
/kent k
This archive was generated by hypermail 2.1.5 : Tue Jan 10 2006 - 16:39:50 CST