On Thu, Jan 25, 2001 at 10:11:35PM +0900, Tomohiro KUBOTA wrote:
> Thus, backspace always erases one *column* regardless of the type of
> the last character (normal, doublewidth, or combined). This de-facto
> standard should also be taken by readline library and so on.
But . . . it's broken. The only way you would could rationalize that is
by reference to the internals of older character sets and terminals. It
also doesn't extend to anything that doesn't have the same column concept
as a terminal. In a proportional system, should U+2230 take two or three
backspaces to delete, or should it take one, despite being wider than all
those 'doublewidth' characters. How many backspaces should it take to
delete 6 Six-per-em Spaces (U+2006)? I would seriously regret losing that
consistency between terminals and wordprocessors for a purely legacy
reason.
Personally, I think it's better ignore this and force a few people to
retrain their fingers rather than add cruft to programs and the people
that use them.
> Note that Linux is likely to run only with GNU libc and we can expect
> these functions are always available.
I think the kernel people would have fits at (a) having the kernel
depend on GNU libc and (b) trying to use libc from inside the kernel.
They're both cases where you lose a lot of flexibility (with (b) being
very bug-prone), and I don't think the kernel people will find the
tradeoff acceptable.
--
David Starner - [EMAIL PROTECTED]
Recovering from a hard drive "crash" - website down
-
Linux-UTF8: i18n of Linux on all levels
Archive: http://mail.nl.linux.org/lists/
