On Thu, Dec 09, 2010 at 07:25:11PM +0200, Amit Ramon wrote:
> Nicolas Williams <nicolas.willi...@oracle.com> [2010-12-08 13:25 -0600]:
>
> >On Wed, Dec 08, 2010 at 09:17:02PM +0200, Amit Ramon wrote:
> >>Chip Camden <sterl...@camdensoftware.com> [2010-12-08 11:01 -0800]:
> >>>On a related topic, is there any way to get mutt to display RTL for
> >>>certain characters? The Hebrew characters in your signature, for
> >>>instance, are displayed LTR in my mutt, so they read backwards.
> >>
> >>Not directly. For that you have to use a bidi-aware terminal. I'm
> >>running Mutt in Mlterm terminal emulator, which can handle LTR and RTL
> >>quiet well (but not perfectly).
> >
> >I agree, this is a job for the rendered, which is also why you shouldn't
> >need your plain2html program -- the web browsers displaying your email
> >in webmail apps should handle bi-di correctly as long as they understand
> >UTF-8. Might the webmail backend be doing something wrong?
>
> I don't agree... web browsers are not supposed to be able to know how
> to render bidi text. For this reason there are dir tags in
> HTML. Webmail backends are also not doing it. This is not the same as
> "understand UTF-8".
I'm not too familiar with bi-di, and I can see that a dir tag does exist
for HTML. We seem to agree that terminals (which have no HTML-like
tags) are supposed to figure out how to render bi-di correctly. Looking
around a bit I see that there are three standard ways to indicate
direction changes:
- Use Unicode control characters, most of which are discouraged, except
for the right-to-left and left-to-right marks (which are for
specifying direction for weak-directional characters relative to
surrounding strong-directional characters);
- Use HTML dir attribute or bdo element;
- Use CSS rules ('direction' and 'bidi-override' props).
The last two are for HTML docs only. The first one is the only one that
works in all contexts, while markup solutions based on anything other
than Unicode require tags/attributes to be defined.
But there is a Unicode bi-di algorithm... From what I can tell,
renderers that implement it should not require marks (except for
weak-directional characters, as mentioned above).
> >How would I know if the Hebrew text in your signature is displaying
> >correctly? [...]
>
> The first Hebrew letter in my first name is "Ayn", which looks
> schematically like this:
>
> [...]
>
> In a correct appearance you should see it at the rightmost place on
> the line that has my first name, Amit, in my signature... if it
> follows the word Amit immediately after the blank space, the terminal
> does not support bidi.
Indeed, my terminal is not displaying those in right-to-left. I see
that some applications do display properly (for example, the Bluefish
editor does, but vim/gvim does not).
Nico
--
