On Tue, 06 Nov 2007 11:33:23 +0100
Abdelrazak Younes <[EMAIL PROTECTED]> wrote:
> Martin Vermeer wrote:
> > On Tue, 06 Nov 2007 08:47:44 +0100
> > Abdelrazak Younes <[EMAIL PROTECTED]> wrote:
> >
> >> Martin Vermeer wrote:
> >>> On Tue, Nov 06, 2007 at 01:39:40AM +0200, Dov Feldstern wrote:
> >>>> Martin Vermeer wrote:
> >>>>> On Mon, Nov 05, 2007 at 11:13:34PM +0200, Dov Feldstern wrote:
> >
> > ...
> >
> >>> Actually I think you should never have any reason to _want_ to do this.
> >>> It's elegant, yes, and I agree philosophically with Abdel. But _in
> >>> practice_ the user can already output whatever he wants, by alternating
> >>> pieces of ERT with non-ERT. Precisely because ERT is "naked", and you
> >>> can set the language attribute for the ERT inset location in the
> >>> containing text too.
> >>>
> >>> I believe ERT (as opposed to, e.g., Listings) should indeed be "naked"
> >>> and never used for anything but outputting TeX-executable stuff.
> >> Well, I agree _pragmatically_ with you ;-) but how does that work in
> >> practice? I agree we should not allow two different encodings nor two
> >> different language within a same ERT. But imagine you want to input an
> >> hebrew word in ERT;
> >
> > Why? Split the ERT in two and input the Hebrew word as ordinary lyxtext
> > inbetween. (I'm sure you have used the construct
> >
> > ERT[\command{]lyxtext text text ... text ERT[}]
> >
> > more than once.)
>
> Ah right, forgot about that :-)
>
> >
> >> then you want to see it display correctly within LyX
> >> and that is RTL. I proposed sometimes ago that the text direction (at
> >> least within a word) should be decided upon the unicode code point only,
> >> not upon the language; I still believe this is the right thing to do.
> >> But right now how can we make the hebrew word display in RTL? I see two
> >> solution:
> >> 1) The ERT inherit the language of the surrounding text: this seems like
> >> counter to the idea that ERT should be naked.
> >> 2) We provide a dialog (or context menu) for ERT to enforce (optionally)
> >> the text direction within the ERT. I actually think this last solution
> >> should be generalized to all text insets.
> >>
> >> What do you think?
> >>
> >> Abdel.
> >
> > Don't make it so difficult ;-)
>
> Right...
>
> Still, I maintain that detecting the text direction automatically based
> on the Unicode code point would simplify things.
Yes, why not actually.
- Martin
