From: Asmus Freytag (firstname.lastname@example.org)
Date: Thu Oct 16 2003 - 00:30:01 CST
I noticed that this message had not gotten a reply.
At 05:07 PM 10/7/03 +0200, Kent Karlsson wrote:
> > >A question about the issues already open: What is the justification
> > >proposing to make Braille Lo?
>Shortly before this came up as a "Public Review Issue", I suggested that
>Braille characters should not be regarded as ignorable symbols when
>collating texts. I.e. that they should have "level one" weights in the
>default weighting table. The reason being that they are more often
>used for letters than for other things. (However, I did not ask to make
That reasoning doesn't really apply. When data-streams contain a mixture
of Braille and other character codes, then one would assume that the Braille
is merely cited for the sighted, in which case it's used as a symbol.
When data streams contain exclusively Braille, then they are actually used
as intended. To sort such data would require a tailoring based on the
Braille mapping being used.
Which you recognize implicitly:
>That would be for the default ordering. Wanting a more
>"alphabetically proper" ordering, would still require tailoring for that
>particular correspondence between "ordinary" letters and Braille, but
>require converting to the "ordinary" letters. Each such tailoring would
>give level 1 weights to most of the Braille characters used in that
That may be true, but I still don't see what difference a change in default
mapping would offer. For people reading the braille, it provides a random,
almost binary ordering, which would seem to swamp all benefits of it being
level 1. [If all data to be sorted has weights only on the same level(s),
the results should be unaffected by which those levels are. Or am I missing
For people using data with Braille embedded, e.g. instructional material,
I don't see the benefit of sorting the Braille as if it was letters by default.
If you wanted to sort a list, like a Braille to English phrasebook, then
you would need a tailored sort anyway.
This archive was generated by hypermail 2.1.5 : Thu Jan 18 2007 - 15:54:24 CST