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

From: Juliusz Chroboczek (
Date: Tue Aug 03 1999 - 06:50:56 EDT

Markus Kuhn:

MK> Note that U+2019, the proper English left quotation mark, doesn't
MK> look like the old GRAVE ACCENT, so we have to use the "phantasy"
MK> character U+201b instead, which is probably not really used in
MK> proper English typography!

The use of these `quotation marks' is merely a convention. As you
mention, it is not typographically correct for English, let alone
French (<<citation>>), German or Polish (,,cytat''). It nonetheless
remains a useful convention for plain text display of those languages.

As a side note, you'll notice that I write

  ``Foo'', he said.

rather than

  ``Foo,'' he said.

Again, this is not proper English typography; it is part of the
convention for quotation marks in plain text.

MK> the comments of people here on the Unicode mailing list (with an
MK> extremely high level of competence in both the history of
MK> character sets as well as typography),

A number of people on this list are very competent in writing systems,
character encoding and typography. This, however, is an issue of
usability and smooth transition on Unix-like systems. Making the
transition from locale-based encodings towards Unicode requires a
subtle balancing act between keeping existing applications from
breaking and making the (future) support for Unicode as good as

In this case, the said competent people on this list are all in favour
of breaking the existing usage as much as possible in order to make
the Unicode support better. I understand their point; their agenda
(pushing Unicode and high-quality text-processing), however, is
somewhat different from mine (managing to provide decent support for
Unicode in X11/XFree86 while not breaking any existing applications).

This is an issue, Markus, on the subject of which we have found
ourselves in disagreement before (on the XFree86 internal mailing
lists, when discussing selections and interaction with locale-based
clients). Unlike some other systems [1], XFree86 needs to be very
conservative in the changes it makes. We cannot ask people to make
such a simple change as upgrade their shared libraries (DLLs), let
alone ask them to upgrade their applications (for one, a large number
of users use X servers from XFree86 for accessing applications on
commercial hosts.) Any change we make in order to support Unicode
must leave the support for locale-based applications unchanged. (And
by this I do not mean ``applications should still work''; I mean,
``all correct applications must remain absolutely unchanged''.)

Note that this is not to say that I am not willing to pay a price for
the Unicode support. As a matter of fact, I have just modified the
current alpha of XFree86 to use significantly less memory for sparse
fonts, such as Unicode-encoded fonts, at the cost of some overhead
with dense fonts. This change was felt to be adequate because the
behaviour of the server with locale-specific fonts has remained
absolutely unchanged, and Mark Vojkovich and I agreed that the
performance overhead is tolerable (it turned out to be completely
negligible in the end).

MK> In fact, XFree86 president David Dawes himself encouraged me right
MK> in the beginning of this project not be be afraid of fixing the
MK> characters 0..255 if I find serious problems with them.

The change you have made in this case significantly breaks the
appearance of a large number of common applications. If David had
such radical changes in mind, then I find myself in disagreement with
him for the very first time.

MK> I am hoping on your support in this.

Your hope is misplaced.

MK> We should identify those GNU packages that currently use `quote'
MK> in output messages and submit small patches that substitute this
MK> usage for 'quote'.

You may try; I fear, however, that you will get as little support from
the maintainers of these packages as you are going to get from me.
(Please do not take this sentence as a dare; contrariwise, I am asking
you to rethink your position before starting to work on this.)

On the other hand, I would support modifying the applications to
(i) support the use of Unicode-based locales, and (ii) use culturally
adequate quotation marks when a Unicode font is available. I know
that you have done some remarkable propaganda work towards (i) (anyone
who hasn't done so yet should have a look at Markus' web pages), and I
would suggest that you add an entry about (ii) to your Unicode FAQ.



[1] Systems running on a single machine, on which the whole system,
from the kernel to the window system to the user-interface libraries,
comes from a single source. This includes both Windows and the Mac.

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