Abdelrazak Younes wrote:
Elazar Leibovich wrote:
Well, have your words for lunch dude. Bon apetite.
In your example all English words are separated by neutral characters
like ," or space. In MS word, one controls the directionality of the
neutral characters by the input language. So that typing your text is
trivially translating from Lyx's
ENGLISH<change language through F12>, <change lang>ENGLISH
to:
ENGLISH<change language through alt-shift>, <change lang>ENGLISH
Try that at work. It's extremely easy to reproduce.
The bad thing about it, is that setting directionality to space, makes
it different from other spaces, and thus causing a great confusion. (I
once imported English plaintext to MS word, which made the
directionality of all spaces to Rtl, while the entire text was LtR. I
had no idea how all this funny things happened, and it was a challenge
to fix).
The same disadvantages goes for Lyx!
FWIW I agree with you Elazar but this is no surprise as Dov said ;-)
For me there shouldn't be any need to change the language to properly
edit and typeset a mixed RTL-LTR document. AFAIS, there is abolutely no
need to write Hebrew in LTR mode and there is no need to write Latin
based languages in RTL mode. So IMHO, we should do that automatically
based on the unicode codepoint. I am sure this is achievable and that
the user experience will be great.
:) . Remember, LaTeX must have the language information (but that could,
as you say, be achieved from the codepoints; although as Andre' pointed
out, this will not work for all languages, so you'd need to start having
two parallel mechanisms for language determination).
Secondly, you must have the option for override --- even if only just
for the neutrals. I think Elazar will agree with that after the
"challenge" ;) I sent.
But this is way too late to change anything in 1.5. I propose to
implement that (optionally), together with visual navigation (also
optionally) in 1.6.
+1
But yes, let's continue this in a few weeks.
Abdel.
Dov
