Re: Int. letters & is_basic

Re: Int. letters & is_basic

[Robert Dewar]

Tore asks

"I'm wondering about is_basic and to_basic in ada.character.handling.
It seems like the point is to take one letter with some kind of marking,
and turn it into the letter without the marking (e.g. in Norwegian we
use a_grave and e_acute, but they are not letters in the alphabet).
If this is the point, then they are not defined correctly (at least not
for the Norwegian letters o_oblique_stroke and a_ring which are letters
in the alphabet). As I wrote in another posting, I'm making a package
for codepage 850. I'm also planning to make cp850.handling. If is_basic
and to_basic isn't supposed to do what I indicate  above, then I'll
probably make something that does what I want :-). Anyway, I would
like to know if there are other letters in latin_1 which is treated
wrong by is_basic. My main concern is sorting. All comments are welcome!"

Sorry, Basic and To_Basic turn A-ring into A. That's the way it is defined.
Yes, I know perfectly well that A-ring has nothing to do with A, but that's
still the way it is defined. You will need to define your own mappings that
work sensibly :-)

Re: Int. letters & is_basic

[Robert Dewar]

Norman says

"letter with any diacritical mark removed.  Therefore, A.3.2(27) is
incorrect not to have listed a_ring and o_oblique_stroke among the basic
[letters]"

No, Norman is wrong here. We discussed this in WG9 specifically. We knew
that A-ring is a "real" letter. However, the decision as to what are
diacritical marks and what are not is indeed not uniform across the
users of Latin-1. The definition of Basic and To_Basic are based on a
simple rule derived from the names used in the ISO standard. They are
well defined, but maybe not what is needed in some particular national
contexts.

Norman, this specific point was discussed at the CRG meeting held at
the Villars meeting. We all understood the problem perfectly well, but
there is really no other possible solution than that taken in the RM
that would hold water. Personally I wanted to remove Basic and To_Basic
compleely, but could not get agreement on this.

P.S. CLA readers may be amused by Jean Ichbiah's "crossword" test here.The question
is whether different versions of a letter can be used for the down and
across clues. The answer for E-Acute in French is for example yes, the
answer for A-Ring and A in Swedish or N and N-tilde in Spanish is no.

Re: Int. letters & is_basic

[Robert I. Eachus]

In article <4g2pdc$15lc@watnews1.watson.ibm.com> ncohen@watson.ibm.com (Norman H. Cohen) writes:

  > No, he is saying that the ring in a_ring and the oblique stroke in
  > o_oblique_stroke are not diacritical marks, any more than the diagonal in
  > "R" is a diacritical mark placed on a "P".  They are distinct Norwegian
  > letters.  As far as I know, there is no language in which the ring and
  > the oblique stroke occur as diacritical marks.

  They are treated as such in English!  And, yes, both angstrom and
oersted occur in my English dictionary as units of measure. (Excuse
the spellings, I don't want to have to create a MIME message and post
it.)

  Someone with the OED on disk might try checking to see if there are
ANY letters in the Latin-1 set which don't occur there.  There are,
of course, letters in English and the OED that don't occur in the
Latin-1 set.  The most obvious of these is the OE dipthong found in
encyclopedia.  (It was there in earlier drafts but was replaced by
Multiplication_Sign and Division_Sign.  If you care. ;-)

  > This can and should be fixed by a binding interpretation of the ARG.

   This can and should be fixed only by providing culture specific
packages.

--

					Robert I. Eachus

with Standard_Disclaimer;
use  Standard_Disclaimer;
function Message (Text: in Clever_Ideas) return Better_Ideas is...