From: Kent Karlsson (email@example.com)
Date: Fri Dec 07 2007 - 13:20:10 CST
Peter Constable quoted and wrote:
> > I did (long ago) suggest to the UTC, in a beta comment for 5.0 IIRC,
> > that they be mirrored. However, the UTC decided to mirror quote marks
> > instead. The latter I did NOT suggest, and that change was a really bad
> > move, and has since been backed on. However, I still think the ornate
> > parentheses should be mirrored, especially if they are used both in
> > LTR and RTL texts rather than only in RTL texts.
> Since you thought changing the mirroring behaviour of
> quotation marks was a really bad move (and you were right
> about that),
There were many things wrong with that mirroring of quote marks:
1) It would affect very very many (RTL) documents, many of which
are old and unlikely to be maintained. (Including documents in
other encodings than Unicode ones, since these characters exist
in several non-Unicode encodings, including "bidi ones" like
CP1255 and CP1256.)
2) The mirroring as reflected in BidiMirroring.txt was all
wrong, not indicating the proper mirror characters.
3) Mirroring (actually mirroring the glyph) of the low quote
marks, which would have been THE way of mirroring those
characters, produces glyphs that as far as I know are NEVER
used in any context.
> I'm surprised you'd think that changing the
> mirroring property of these characters would be a good move.
1) These characters were relatively recently encoded in Unicode,
and there is likely not very many documents using the ornate
parentheses in a Unicode encoding. Furthermore, they are not
present in commonly used non-Unicode encodings.
2) The font support for these characters is still relatively
weak (few fonts have them), which supports 1.
3) If used in documents (Unicode encoded), it must be a relatively
new use, in the sense that the document are more likely to be
4) Leaving them as is gives strange results for line breaking
(as you noted), and maybe other things too.
5) One of the arguments for not changing the mirroring of
them (but should have swapped the open/close properties), was
that they are ONLY used for RTL text, never for LTR text.
Which apparently is not true; hence they should work in
the same way as ordinary parentheses to w.r.t. mirroring
and line breaking (etc.).
6) Of course it would have been better to have gotten this
"right" early on (and I did give it as a beta 5.0 comment),
but mirroring is not fully stabilised, and can be changed.
In this case I think the positives of a change here still far
outweigh the negatives (which was VERY much NOT the case for
quote marks and a mirroring change).
This archive was generated by hypermail 2.1.5 : Fri Dec 07 2007 - 13:22:54 CST