Hi Jonathan,

On 08.10.2004 13:57, Jonathan Ben Avraham wrote:

a) Do it the MS way (aka the character attribute solution). We would
introduce a character attribute which would automatically be set
depending on the IME which is used to enter the characters. The
information obtained from the character attribute can
be used to resolve the neutral characters. (There is an optional
UBiDiLevel* embeddingLevels parameter in the ubidi_setPara(...) function
call).


How would you set this attribute when there is no IME - like on my RH7.3
Linux box using twm window manager and a keyboard mapping that I
downloaded from Pango org and then hacked?

you are absolutely right. As Falko already mentioned, that's one of the major drawbacks with this solution.



Do we really need to change the OOo file format for this? I don't beloeve so.

If this *would* be the solution we all agree upon, this has to be implemented in the file format as well. No problem with that. There will always be new attributes if we implement new features.


b) Build our own enhancements on top of the Unicode Bidi Algorithm,
e.g., to resolve the f(x) problem as stated in issue 18024. That would
mean we identify a couple of problematic cases and automatically insert
control characters where necessary. This is what mehlng proposes in


Not necessary. No "enhancements" to the standard needed. Just let the user
decide by pressing a shortcut key for LRO, and then PDO when returning to
RTL text flow within an RTL para.

Yes, but that's not what mehlng suggests in i18024. Believe me, I would be happy with just offering the user to insert his control characters manually. Maybe we should see this as our third option: Simply providing some ui to insert control characters without any additional, automatic magic. I'm totally fine with that. We would have to enhance our Word import filter to insert control characters if necessary, maybe we should also visualize the control characters in a special way, and that's it. Keep it simple.


c) Somehow implement both options. Obviously some prefer option a,
others prefer option b. A possible solution *could* be to have the
option "evaluate direction character attribute" and the option
"automatically insert control characters on certain situations". We
would generate the character attribute depending on the IME information,


There's no IME information on my computer. Why try to second guess the
user? There will always be some cases when we guess wrong and that's super
annoying when you intended one thing and the software "corrects" you.

I fully agree.

Have a nice day,

Frank

==============================================================
To unsubscribe, send mail to [EMAIL PROTECTED]
with "unsubscribe hebrew" in the message body.



Reply via email to