Re: [emacs-bidi] debugging visual-to-logical

[Behdad Esfahbod]

On Tue, 20 Nov 2001, Alex Schroeder wrote:

> Personally I don't think it makes much sense to compare things if some
> implementations use different tables.  Is there a really good reason
> to switch from the bidi types assigned in UnicodeData.txt to the one
> the Reference Implementation is using?  If there is, I will change the
> tables.  I'd like to see a URL with the new definitions, though, so
> that I can trust it's legitimacy.  If there is no really good reason
> to use bidi types differing from UnicodeData.txt, however, I think we
> should all make it perfectly clear which tables we are using and why
> we think it is relevant in our post.  Because if we don't do that, we
> will be confusing ourselves to no end.  I was quite confused when I
> read this matter-of-fact statement:
> 
> >> > No, its ".3210-555-02 :SI REBMUN YM".
> 
> Just to make myself clear, my bidi.el uses the bidi types from
> UnicodeData.txt whenever I am presenting code or discussing questions.
> The only exceptions are the uppercase letters [A-Z] which use type R
> instead of L.

You are right, quoting from fribidi's description of CapRTL
('fribidi --charsetdesc CapRTL'):

        CapRTL is a character set for testing with the reference
        implementation, with explicit marks escape strings, and
        the property that contains all unicode character types in
        ASCII range 0-127.

and the Reference uses this types, to let you test all Unicode types 
just with ascii strings, much like the reason you use A-Z fro R, but 
the are other types there too, like AL and AN.

My opinion is to use the CapRTL encoding everywhere that you want 
something different than Unicode's types (when you assign somthing 
different than L to A-Z), BTW, the Reference's character types is 
changed in the newly released code, but I refer to the code that is 
include in fribidi ('fribidi/conformance/reference/bidiref.cpp'), you 
can find the its character types from its source, or by running 
'fribidi --charsetdesc CapRTL'.

-- 
Behdad
30 Aban 1380, 2001 Nov 21

[Finger for Geek Code]