Re: [emacs-bidi] adding LRM in visual-to-logical transform

[Behdad Esfahbod]

On Tue, 20 Nov 2001, Eli Zaretskii wrote:

> I'm not sure I understand.  Are you saying that these two
> logical-order strings are ``equal''?
> 
>    he said "I NEED WATER!"
>    he said "I NEED WATER!<RLM>"

Yes! see below.

> Perhaps you meant removing the directional marks after resolving the
> explicit embeddings?  But if so, the embedding resolution leaves
> behind information that is not part of the string itself, so
> comparing the strings alone after removing the marks won't tell the
> whole story.
> 
> What am I missing?

Well, I have some points:

1. Logically the two tests above are the same: you read them the same, 
they have the same meaning: the word WATER is followed by the bang, in 
both bang is in the quotaion marks, then both mean that the bang is 
related to the inner sentence, independent of how you display it.

2. In Unicode standard, there is more that one equality meaning:
* Case insensitive, * Accent insensitive, * Weak, * Strong ...
and almost in all of them put away the directional marks before doing 
any comarisions.

3. There is no problem with Explicit Overrides and Embeddings: the 
text is in *always* in logical order, the bidi marks are just there to 
help renderer show the visual order correct, if you put away the bidi 
marks and write the string (in logical order) on the screen, and read 
it from left to right (Yes, read hebrew and arabic from left to 
right), then you find that any to strings that you expect to be equal 
are the same here, and reversely any two that look to be equal, have 
the same meaning.

4. Consider the logical order as an audio stream, which someone is 
reading, and you type it in your computer, it's unique up to here, 
then you add bidi marks to make it look nicer, then if two strings are 
equal (semantically equall), then removing the bidi marks, they should 
be equal byte by byte.

-- 
Behdad
29 Aban 1380, 2001 Nov 20

[Finger for Geek Code]