23AF HORIZONTAL LINE EXTENSION: glyph or variation selector?

Ilya Zakharevich nospam-abuse at ilyaz.org
Wed Apr 2 14:35:28 CDT 2014

On Wed, Apr 02, 2014 at 06:56:54PM +0000, Whistler, Ken wrote:
> Ilya,
> U+23AF is *definitely* not a variation selector at all.
> It is part of a set of bracket pieces (and other graphic pieces)
> in the range U+239B..U+23B1.

Evidence does not support this (see below).

> See discussion of the topic at:
> http://www.unicode.org/forum/viewtopic.php?f=35&t=206

Apparently, this has no relationship to U+23af at all…

> See also Section 2.13 of UTR #25:
> http://www.unicode.org/reports/tr25/


> which discusses the use of these symbol pieces. It does not
> specifically talk about the arrow extender pieces, focusing
> instead on the bracket pieces, but the principles are the same.

AFAIU, they are not.

> These glyphic pieces of symbols are only relevant and useful
> in the context of mathematical typesetting programs like TeX.

Are you sure?  Did you look at Figure 6 in Appendix F of TeXBook?

  • There is no horizontal extension pieces;
  • The vertical extension pieces consists of very short chunks (to
    make size tunable in small increments).

The horizontal extension of arrows in TeX is done by macros like
\longleftarrow and \Longleftarrow (Appendix B, or just look in
plain.tex).  However, these macros (again!) have no relation to
U+23af, since they need DIFFERENT extension pieces for
single/double/triple/etc arrows.


In short: if U+23AF were a part of extensible set, it would be short
(never saw it like this) AND would have double/triple/etc

> The set of box drawing characters in the U+2500 block were
> encoded for compatibility with old character sets that did
> character cell graphics.
> So the two are different, but neither set is of much current
> relevance for general text currently using arrows.

I won’t be so sure.  I use extended arrows in “general text” mode;
they are not shown reliably in ANY environment I know.  I’d love to
know a solution.


More information about the Unicode mailing list