Re: plain text, paragraphs, and bidi

[Roozbeh Pournader <roozbeh@xxxxxxxxxx>]

On Wed, 17 Jan 2001, Bruno Haible wrote:

> We can request that all explicit (stateful) bidi marks must be
> terminated before the end-of-line. Cut&paste and line breaking
> algorithms will have to be careful.

What about also having an extended mechanism for cutting and pasting that
works also in paragraph level? If there are two tools that respect
the traditional paragraph separated with empty lines, or their own notion
of paragraph that's better than our plain vanilla idea, they should have a
way to cut and paste into each other.

> Which is not really better. The real solution is probably to use
> implicit directional marks (RLM, LRM) only.
> 
> Stateful bidi marks are like stateful encodings (ISO-2022), and they
> have the same problems: they add complexity to simple tasks like line
> breaking and selecting/extracting a piece of a line.

They add to the complexity, I agree, but they are more than needed for
many text encoding applications. LRM and RLM are just not enough, since
there are many cases were while the visual text should come out in a
particular order to be readable by the reader, the logical text should
also have another particular order to make it processable, and cases arise
in places as simple as section numbers or dates.

Even with current practice of Unicode bidi, and not considering legacy
tools behaviour, we (bidi writers) have a really hard time even editting
normal text. And we also have a lot of documents in mousetraps: the
software does not provide a spec for the file format, and the text is more
complex (only using numbers and symbols) than can be allowed to converted
incorrectly. In short, we even do not even have portable plain text. :(

--roozbeh


-
Linux-UTF8:   i18n of Linux on all levels
Archive:      http://mail.nl.linux.org/lists/