Emacs & Hebrew

[Eli Zaretskii]

> From: Omer Zak <w1 at zak.co.il>
> Cc: Eli Zaretskii <eliz at gnu.org>, linux-il at cs.huji.ac.il
> Date: Fri, 15 Jun 2012 10:55:13 +0300
> 
> Nice example!
> 
> It could be complicated further by having strings with mixture of LTR
> and RTL glyphs.  Such as:
> 
> string.tr("abcdefghijklmnopqrstuvwxyz"
>           "ABC,EFG abc: HIJ>KL d<e MN")
> // small characters and LTR and capitals are RTL.
> 
> In which BiDi reordering would leave the software developer very
> confused if he wants to figure out into which glyphs do 'q' and 'v' get
> translated.  Or which glyph was the original for ':'.

We could have specially-formatted comments, or some other way, to tell
the display engine not to reorder a certain portion of the source
code.  Once selective reordering is available, it would be easy enough
to use it in such ways to cover special cases.  We already have
similar features via file-local variables, 'coding' tags, etc.

The main point is that we should not punish 90% of use cases for the
sake of the other 10%.  Holding a feature because it only handles 90%
of use cases is a bad call, IMO.

> My preference, when editing software, is to show all text in logical
> order, but provide the user with a popup window showing the visual
> order.

This will make reading a source with a lot of Hebrew text extremely
cumbersome, to say the least.  Again, the "usual" cases should "just
work".