Re: less & composing characters - fix

[Behdad Esfahbod <behdad at cs dot toronto dot edu>]

On Mon, 23 Feb 2004, Mark Nudelman wrote:

> >   * About Non-Spacing Marks like Fatha it makes more sense.  But
> > again, the problem is that:  Terminals like xterm and probably
> > mlterm, put the mark on top of the character, so no width
> > allocated for the mark.  But what we're gonna do in Linux
> > console?  There (as done by BiCon now), a non-spacing mark still
> > occupies a single width.  So how's less gonna handle this case?
>
> Oh dear.  This is news to me.  So there are some terminals which honor
> composing characters (make them take zero width), and others that do
> not?  Is it safe to assume there are only two types of terminals, those
> that honor all composing chars and those that honor none of them?  Or

I'm afraid not.

> could there be a terminal that treats some composing chars as zero width
> and others as normal spacing chars.

Well, I really don't like the idea of less handling this kind of
things :(.  It is quite probable that people run a couple of
filters on their terminal.  More advanced terminals like mlterm
can be assumed as filters themselves.  This filters may handle
composing characters in any random way.

For example, in console, a Latin non-spacing mark is better
ignored if a suitable glyph is not available for it put on the
base glyph, but an Arabic one should not.

I strongly recommend that you ask for terminal experts' opinion
on linux-utf8 mailing list.

behdad

> --Mark