Re: Latin-1's apostrophe, grave accent, acute accent

From: Markus Kuhn (Markus.Kuhn@cl.cam.ac.uk)
Date: Tue Aug 03 1999 - 04:33:25 EDT


Juliusz Chroboczek wrote on 1999-08-03 06:54 UTC:
> I would like to remind you that you are not only pushing for your
> fonts to be included in XFree86; you are actually pushing for the
> fonts to be the default `fixed' fonts.

I am perfectly aware of that.

> Whatever change you make to codepoints 0 up to 255 will have an effect
> not only on Unicode-based applications, but also on all applications
> using locale-based charsets when running in an ISO 8859-1 locale. In
> these applications, the Unicode characters for quotes will not be
> accessible. At all.

I am also perfectly aware of that and believe me, I have very seriously
considered this aspect.

[Note by the way, that we can provide in X with your new font backend
recoding mechanisms a special encoding *-iso8859-x.oldstyle (or
whatever) which maps 0x27 and 0x60 onto U+201b/U+2018 for those - like
yourself - who need a little bit more time to get over the ASCII
directed-quotation withdrawl symtoms. I will make sure, that all fixed
fonts will include U+201b. Note that U+2019, the proper English left
quotation mark, doesn't look like the old GRAVE ACCENT, so we have to
use the "phantasy" character U+201b instead, which is probably not
really used in proper English typography!]

> There currently is a large number of Unix applications that use
> `\verb|'|' and `\verb|'|' as single quotes. As Roozbeh noted, this
> includes TeX, but also makeinfo (hardwired) and m4 (customizable).
> Furthermore, many users (including me) have their fingers hardwired to
> using them when quoting.
>
> I would like to very strongly ask you to revert to a slanted glyph
> shape for the apostrophe. Changing this will only become practical
> when most Unix applications are Unicode based.

No, please understand that this is *not* only a Unicode related problem!
This is has been a problem of the X11 fonts for a long time that has
been a cause of confusion completely independent of Unicode. It is in
fact a small incompatibility between US-ASCII and Latin-1 caused by the
addition of ACUTE ACCENT to Latin-1. Unix users have been mislead by the
design of the X11 fonts for a long time into believing that the way they
see APOSTROPHE and GRAVE ACCENT will look comparable on all available
output devices. This is most certainly not the case! Users on other
platforms who write feet and inches as 6'3" currently get a funny
looking result on X11 screens (I just didn't mention this example,
because the real solution is in my opinion to use SI units instead :-),
and users who cut&paste email excerpts into most word processing
packages see similarly strange results for `quote'. The mess up with the
accent keys on e.g. the German keyboard is a direct consequence of the
old glyph shapes. This should have been fixed already back in 1991 or so
when the X11 fonts were extended to Latin-1 and there became a proper
code for ACUTE ACCENT available.

The user feedback that I received so far on this change in the fixed
fonts centered just around the fact that I hadn't made the change yet in
all fonts, and therefore people complained about a temporary
inconsistency in the various font sizes, but not about the change in
general.

Note also that GRAVE ACCENT is really the wrong sort of quotation mark
anyway. In English, the left quotation mark (U+2018) is a raised comma,
rotated by 180°, while the old GRAVE ACCENT looks more like a mirrored
comma (U+201B), which is normally not used in English typography. So
since it was the wrong shape anyway, replacing it by another shape that
also is not a very correct left quotation mark doesn't matter *that*
much. GRAVE ACCENT just remains the wrong glyph for an English left
quotation mark.

I have hesitated about this change for some time, for exactly the
(rather few) reasons you gave, but not only the comments of people here
on the Unicode mailing list (with an extremely high level of competence
in both the history of character sets as well as typography), but also
first user feedback has strongly encouraged me to go ahead. In fact,
XFree86 president David Dawes himself encouraged me right in the
beginning of this project not be be afraid of fixing the characters
0..255 if I find serious problems with them.

There are only two widely used Unix applications that use the GRAVE
ACCENT and that cannot be changed easily. But for both applications, it
fortunatelly does not really matter:

  - In the Unix shell, the GRAVE ACCENT is always used in pairs of
    GRAVE ACCENTs, and not together with an APOSTROPHE on the other side.
    So "echo `ls`" will look as reasonable as before, in fact, the visual
    difference between "echo 'ls'" and "echo `ls`" is now even a bit
    clearer, which slightly increases the readability of shell scripts
    and which will make it less likely that beginners mix up these two
    characters with very different functionality. The fact that the glyph
    shapes now correspond to the common keyboard labeling practice will also
    significantly reduce potential confusion among beginners.

  - In TeX, there are numerous special ligatures used to access typographic
    characters not available in ASCII. E.g. "---" is interpreted as an em-dash
    and "`" and "'" are interpreted as left and right single quotation marks.
    All of these are already somewhat ugly compromises of readability,
    and neither the old nor the new GRAVE ACCENT really looked ever much like
    the U+2018 shape that TeX makes out of it. So the small loss of symmetry
    that the correction of APOSTROPHE and GRAVE causes in TeX input like
    `quote' and ``quote'' doesn't matter that much in an environment
    where plain text is anyway scattered with backslashes and markup
    commands, and where visual accuracy of the plain-text appearance
    was certainly never a strong design goal.

I am hoping on your support in this. We should identify those GNU
packages that currently use `quote' in output messages and submit small
patches that substitute this usage for 'quote'. I'd expect this to be a
very manageable task. I am convinced that the overall user friendliness
of X will in fact be increased by the change of these glyph shapes for
the reasons stated. This is *not* just a matter of blind standards
conformance. It is a real bug fix.

Markus

-- 
Markus G. Kuhn, Computer Laboratory, University of Cambridge, UK
Email: mkuhn at acm.org,  WWW: <http://www.cl.cam.ac.uk/~mgk25/>



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:50 EDT