Title: DS additional comments on P1003.2b/D11

Date: 1995-10-21

Action: For WG15 meeting 1995-10

Danish Standards has the following additional comments on the
P103.2b/D11 document, in excess of previous comments as noted
in WG15 N586 and in our DIS ballot (symbolic ellipses):

@ 2.2 c 1

2.2.3.201 The SC2 name is "UTF-8" with a hyphen. Please refer
to the standard, it has recently been approved as an IS.

@ 2.4 c 2

We do not see what the revised wording accomplishes, what it
says is that you can have names like <Uxxxx> or <Uxxxxxxxx> but
that was already the case. See our comments on the localedef
utility.

@ 2.4 o 3

2.4.1 WIDTH, WIDTH_VARIABLE, WIDTH_DEFAULT statements should be
in the locale definition, as it is not coded related, but
character related. For example the character <double width A>
is present in a number of coded character sets, including IS
10656, JIS X0208, GB2312 and KSC 5601. So this is independent
of coding and a property of an abstract character, and the
right place is then in the locale, possibly in LC_CTYPE. The
WIDTH etc can also be used to represent fonts, and then it
should be the same regardless of the coded character set, for
example an <a> in ASCII or ISO 8859-1 or ISO 8859-2 has the
same properties in the Helvetica font.

@ 2.5 o 4

2.5.2.2.4 - we do not understand why coding of NUL was
retained, as the standard should cater for other languages than
C, and not impose C ideosyncraziees on the other languages.

@ 2.8 o 5

line 379: The range experssion should not be dependent on the
collation element order, but rather the result of the
comparison using the relevant collation. Using the collating
element order is not proper, and confusing to users that only
has expectations as defiend by the collation rules.

@ 4.35 o 6

line 1102: We should use the repertoiremap files as requested
by WG20, Denmark and Canada, and defined by CEN and X/Open.

The repertoiremap files should be used on the localedef command
line to in a well-defined way be able to use locales and
charmaps using diferent symbolic character notations. Two new
options are proposed:

     -l repertoire_name       Specify the name of a
			      repertoiremap for mapping the
			      locale symbolic character names
			      to IS 10646

     -r repertoire_name       Specify the name of a
			      repertoiremap for mapping the
			      charmap symbolic character names
			      to IS 10646

The -u option should be removed as the above covers the same
functionality in a much more well defined way, and as much data
for this is available.

A naming conviention should be introduced to reference standard
locales, charmaps and repertoiremaps from the international
cultural conventions registry, for example introduced by the
three letters "std".

@ 4.48 o 7

Line 2150: the name should be "codeset" not "charset" - we are
talking about coded character sets, not the character sets
which is without coding, acording to SC2 terminology.

Line 2155: plese do not use the term "ISO-IR" as this means ISO
International Registry , as done in ISO 2375, and what you
refer to on the following is not IS 2375 entries. For example
UTF-8 is not in the IS 2375 registry. If you use just "ISO" or
"ISO/IEC" or "IS" it would be fine. We would suggest that you
use a notation as close as possible to the ISO naming, for
example "ISO 8859-1:1987" - not just spaces. It would be
preferrable that such an item could be specified as just one
token in a shell command line, and it is proposed to use
<underscore> for <space>. Alignment with the forthcoming
international registry that includes POSIX charmaps is
requested.

Please use "UTF-8" instead of "UTF8", and please specify
wheather it is UCS2 or UCS4, and possibly also the endianness,
not just ISO 10646.

@ D.1 c 8

Annex D: a new RFC is available in RFC 1521