Michael Everson: Fonts with buailte characters

Thu Apr 24 12:27:18 UTC 2003

MODERATOR NOTE: This note contains comments from Michael Everson who
refutes some comments from Marion Gunn from an earlier Celtling post.
Celtling cannot verify either party's claims.

---------------------------

Date: Wed, 23 Apr 2003 10:14:30 +0430
To: GAELIC Language Bulletin Board <GAELIC-L at LISTSERV.HEANET.IE>,
         gaeilge-b at LISTSERV.HEANET.IE
From: Michael Everson <everson at evertype.com>
Subject: Re: Tom Pullman: Fonts with buailte characters
Cc: CELTLING at LISTSERV.LINGUISTLIST.ORG

At 10:49 +0100 2003-04-19, Marion Gunn wrote:
>>
>>Strictly speaking, it is to ISO (The International Standards
>>Organization) rather than Unicode, credit is due for the initiative
>>of making our dotted consonants into an international standard
>>(Unicode's involvement is at the secondary/implementation level, in
>>agreeing to respect that ISO Standard).
>
This is quite misleading. The Unicode Standard as a character set is
identical to the character set in ISO/IEC 10646. Seen from that point
of view, the two standards are entirely equal. Unicode, however,
*also* provides implementation standards and guidelines, which
ISO/IEC 10646 does *not* -- which means if either of the two
standards is subordinate to the other, it is ISO/IEC 10646. But in
terms of the development of the character set, both standards are on
equal footing, and it should not be suggested otherwise.

ISO/IEC 10646 is indeed vitally important, for public procurement,
for instance, and it is likewise important that it be maintained as
an international standard, so that the ISO/IEC JTC1 National Bodies
and the computer industry both keep checks and balances on each other
as the Universal Character Set is developed. The two committees
responsible for the standard have an excellent working relationship,
and neither side imagines that it is "better" than the other nor
tries to take advantage of the other.

Regarding the dotted consonants, Unicode 1.0 provided for a COMBINING
DOT ABOVE which could be used together with base consonants to
represent Irish letters -- so Unicode has always supported Irish
(though it didn't support the "agus" sign until I proposed it for
encoding!). Way back in 1992, however, it was seen to be advantageous
to ensure that precomposed letters with dots needed for all the Irish
characters were available, and NSAI voted no on DIS 10646, changing
its vote to yes when the missing characters were added. These were
published in the first edition of ISO/IEC 10646 and in Unicode 1.1.
Today, the LETTER B plus COMBINING DOT ABOVE is canonically
equivalent to LETTER B WITH DOT ABOVE, which means you can use either
interchangeably. Precomposed font glyphs for these characters are
undoubtedly useful, and it is the case that many fonts still do not
have them. I continue to work to encourage vendors to support them,
particularly in advocating the use of the MES-2 and Modern European
Scripts subsets. These two subsets were drafted by CEN/TC304 and
finalized in a CEN Workshop Agreement for which I was project editor.
The MES-2 specification can be seen in
http://www.evertype.com/standards/iso10646/pdf/cwa13873.pdf -- the
MES-3 specification there was expanded into the Modern European
Scripts subset by ISO/IEC JTC1/SC2/WG2.

Currently I am discussing the matter with a working group on mobile
telephony. I may not be successful, but I am urging them to support
the whole of the MES-2, which would mean that all the dotted
consonants, not just the Maltese C and G, might be on a new
generation of mobile phones. We shall see.

>>This initiative, which was sponsored, worked on in conjunction with
>>many people in Ireland and elsewhere, all spurred along in
>>international fora by my company (EGT.IE)
>
It appears from this text that Marion would like "her company" to
take the credit for this work, but in fact it was I who drew up the
ISO/IEC work-item proposal, and it was I who was editor of ISO/IEC
8859-14, ushering it from New Work Item proposal to Committee Draft
to Draft International Standard to International Standard. Of course
many people did participate in the work, including members of
NSAI/AGITS/WG6 (now NSAI/ICTSCC/SC4) and members of ISO/IEC
JTC1/SC2/WG3.

It is true that when I did these things I was Marion's business
co-partner; as some readers may know, that business partnership ended
formally on 2001-09-21. But I think that it is a wee bit misleading
to suggest that "EGT" maintains either interest or activity in the
field of international character set standardization. I think that
credit should be given where credit is due, and it is certainly the
case that in the past decade I have, as Irish representative to
ISO/IEC JTC1/SC2, been one of the leading forces driving the encoding
of scripts and characters used by linguistic communities both large
and small worldwide. That I did this while in business partnership
with Marion for a time is a fact, but I continue to do it on my own.
On a number of occasions during the past year Marion has made similar
revisionist statements about "her company's" participation in this
work on a number of international fora where I am active. In each
case she has *avoided* even the use of my name; I find this
disingenuous, which is why I feel moved to respond to such statements.

>>and NSAI (The National Standards Authority of Ireland) for years,
>>until it was formally defined, to our great joy, in 1998 as ISO
>>8859-14 (aka  Latin-8), and which can be seen in the context of
>>related ISO standards on URL:
>>http://www.eleves.ens.fr:8080/home/madore/computers/unicode/cstab.html#Latin-8
>
The final draft text of 8859-14 can be found at
http://www.evertype.com/standards/iso8859/8859-14-en.pdf -- the
standard was formally published on 1998-11-15.

>>For an account of Unicode's 1999 agreement to implement the
>>provisions of ISO 8859-14 by mapping the characters it defines to
>>their system ('the data the Unicode Consortium has on how ISO/IEC
>>8859-14:1998 characters map into Unicode'), see:
>>http://www.unicode.org/Public/MAPPINGS/ISO8859/8859-14.TXT

That isn't an "account" of an "agreement". It's a mapping table made
by Markus Kuhn and Ken Whistler. It does, however, mean that data
properly tagged as being in 8859-14 can be converted to Unicode/10646
data if one uses conversion software implementing this mapping table.
>
>>For an expert opinion on various implementations in keyboard form, see
>>Ciarán Ó Duibhín's essay:
>>http://www.smo.uhi.ac.uk/~oduibhin/mearchlar/latin8.htm.
>
This page does not discuss keyboard implementations, but rather
discusses the 8859-14 character set and some font extensions to it
for use on Windows. These extensions differ somewhat from those which
I have recommended
(http://www.evertype.com/celtscript/celtcode.html). In particular,
Ciarán and I disagree on the proper use and differentiation of the
AMPERSAND and the TIRONIAN SIGN ET in Irish data.

Regarding keyboards, Ciaran also favours a dead-key apostrophe to
access letters with sínte fada, while many prefer the alt+vowel
solutions devised by Mike Brady and myself in 1989 and 1990
respectively. Additional information on keyboards and the Mac OS can
be found at http://www.evertype.com/celtscript/celt-keys.html
>
>>Cc: GAELIC-L - cuid dár stair, a chairde, nár thuig cuid againn, is
>>dócha, agus sinn ag dul tríd, is nárbh éasca a mhíniú/thuiscint ag
>>an am. Tá súil agam go léiríonn an teachtaireacht seo an éifeacht a
>>ghabhann le gníomhú in éineacht, in am trátha.
>
Tá tábhacht sa stair sin, gan dabht, a chairde, ach bheadh sé níos
fearr gan an stair a athinsint agus ainm an ghníomhaire a fhágaint
amach as an scéal. Cuireadh a chuid saineolais ar fáil ar son na
cúise, agus cuirfear freisin amach anseo.
--
Michael Everson * * Everson Typography *  * http://www.evertype.com

--
o.o.o.o.o.o.o.o.o.o

CELTLING
Post: celtling at lists.linguistlist.org OR celtling at listserv.linguistlist.org
Archives: <http://listserv.linguistlist.org/archives/celtling.html>
Subscribe/Unsubscribe - Go to Archives, then click "Join or leave" link

Website: <http://www.personal.psu.edu/ejp10/celtling>