https://bugs.documentfoundation.org/show_bug.cgi?id=172081

Eyal Rozenberg <[email protected]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |needsUXEval
                 CC|                            |[email protected],
                   |                            |libreoffice-ux-advise@lists
                   |                            |.freedesktop.org
           See Also|                            |https://bugs.documentfounda
                   |                            |tion.org/show_bug.cgi?id=17
                   |                            |0766,
                   |                            |https://bugs.documentfounda
                   |                            |tion.org/show_bug.cgi?id=16
                   |                            |2120

--- Comment #2 from Eyal Rozenberg <[email protected]> ---
Now, as for my thoughts about this... 

This is a bit of a pickle. Shem is indicating one 'scenario' in terms of user
intent, while the implementation assumes another:

* Shem's implicit user intent: "I want autodetection. It usually makes the
right call, and sometimes needs to be overridden. In a specific paragraph, I
need to override it - but that doesn't mean I want to turn if off."
* The user intent implicit in the implementation: "The LTR and RTL buttons
apply a direction as direct formatting which sticks. So, once I press them, I
expect the effect of that press to continue until I indicate otherwise. And
pressing Enter does not indicate otherwise - like with all (paragraph) DF."

Both of these perspectives are legitimate. Why do they have to clash?

Well, they are certain to clash with the RTL/LTR buttons also having an effect
on autodetection - and the effect being implicit. And there is also the
question of what it means for a paragraph to have an explicit direction setting
while also having autodetection turned on (and that's also a question regarding
the ODF document model, i.e. what meaning does the value of style:writing-mode
have when style:writing-mode-automatic is in effect).

We should also note that the UI in Format Paragraph is separate, and
corresponds to the ODF model of two separate/independent properties - in
contrast with the Toolbar and Tab Bar commands. And finally, we may also note
that Format Paragraph has an Inherit direction value, which the Toolbar and Tab
Bar also don't expose.

Now, we could:

1. Keep things just as they are, preferring the original rationale.
2. Have the RTL/LTR button click "stick" for the current paragraph, while
autodetection is on.
3. Have a separate direction autodetection button, with the RTL/LTR buttons not
having any effect.
4. Have a three-mode button (like in my Thunderbird extension): Forced RTL,
Forced LTR, Autodetect. The "forcing" would not mean forcing by DF, but rather
any writing-mode setting with a false value of writing-mode-automatic.
5. Have some kind of visual indication which is not a command for autodetection
mode. Say, a highlight around the buttons, or another modification of the image
on them. Then, pressing the buttons removes the highlight. The button behavior
will remain the same, but the user would get at least some indication of there
having been a change.

And there may be other alternatives still.

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to