[news.eclipse.tools.gef] Re: RTL and bidi behavior

["Thuc" <tqnguyen@xxxxxxxxxx>]

You're right that case 1 is showing the phrase correctly. Thinking back on
what I wrote, what I really meant as "correct" behavior is that the RTL
phrase appear properly in RTL mode. Of course with case 1, with everything
set to LTR, the text is displayed using LTR layout.

Once GEF RTL and bidi development is working 100%, is it correct to say the
following?
With the 4 cases below, we will see only two different results. Regardless
of the EditorPart orientation, the text will be one of the following based
only on the BlockFlow orientation:

BLockFlow orientation = LTR
"25 :<rtl text> (USD)"

BLockFlow orientation = RTL
"(USD) 25 :<rtl text>"

thanks

"Pratik Shah" <ppshah@xxxxxxxxxx> wrote in message
news:d1r09o$sm9$1@xxxxxxxxxxxxxxxxxxx
Case 1 is correct behaviour.  I've verified that TextLayout does the same.
"Randy Hudson" <none@xxxxxxxxxx> wrote in message
news:d1pgb8$tv4$1@xxxxxxxxxxxxxxxxxxx
> -----------------------------------
> case 1: incorrect
> EditorPart.getOrientation = LTR
> BlockFlow orientation = LTR
> "25 :<rtl text> (USD)"

Why is this incorrect? should it be:
<rtl text>: 25 (USD)
?

> -----------------------------------
> case 2: correct
> EditorPart.getOrientation = LTR
> BlockFlow orientation = RTL
> "(USD) 25 :<rtl text>"

Good, this is the only case we have tested.

> -----------------------------------
> case 3: incorrect
> EditorPart.getOrientation = RTL
> BlockFlow orientation = LTR
> "(USD) <rtl text> :25"

You are seeing mirrored, LTR text.  Currently we are not expecting mirroring
to be enabled.

>
> -----------------------------------
> case 4: incorrect
> EditorPart.getOrientation = RTL
> BlockFlow orientation = RTL
> "<rtl text> :25) USD("
> -----------------------------------
You are seeing mirrored RTL text.  The text is being drawn in fragments,
which are arranged RTL, but then SWT flip-flops the locations to their
mirrored positions, giving the the above result.  We are working on
supporting this case.

> Case 1 seems to be using LTR text layout rules, treating the colon and
> number 25 as part of the rtl word.
> Case 2 is the only correct scenario.
> Case 3 uses some strange ordering.
> Case 4 text layout is bizarre.
4 should be simple to fix.  I'm still worred about 1 though.  If 1 is
working and 4 is fixed, then 3 will work too.