From eliz at gnu.org  Mon Sep 28 10:17:20 2009
From: eliz at gnu.org (Eli Zaretskii)
Date: Mon, 28 Sep 2009 19:17:20 +0200
Subject: [FriBidi] Implementation of X6 in the FriBidi library
Message-ID: <83ocovq99b.fsf@gnu.org>

Hi,

Sorry if this is a FAQ, and please point me at it if it is.

The file fribidi-bidi.c in the FriBidi library has this code:

    for_run_list (pp, main_run_list)
    {
      FriBidiCharType this_type = RL_TYPE (pp);
      if (FRIBIDI_IS_EXPLICIT_OR_BN (this_type))
	{
	  ...
	}
      else
	{
	  /* X6. For all types besides RLE, LRE, RLO, LRO, and PDF:
	     a. Set the level of the current character to the current
	     embedding level.
	     b. Whenever the directional override status is not neutral,
	     reset the current character type to the directional override
	     status. */
	  RL_LEVEL (pp) = level;
	  if (!FRIBIDI_IS_NEUTRAL (override))
	    RL_TYPE (pp) = override;
	}

Unless I'm misreading the code, this means that X6 is not applied to
characters whose original type is "WEAK BN".  However, UAX#9 does not
exempt WEAK BN characters from the override, so this looks like being
against the UAX#9 rules.

Am I missing something? are there good reasons for behaving like that?

TIA

From behdad at behdad.org  Mon Sep 28 12:58:12 2009
From: behdad at behdad.org (Behdad Esfahbod)
Date: Mon, 28 Sep 2009 15:58:12 -0400
Subject: [FriBidi] Implementation of X6 in the FriBidi library
In-Reply-To: <83ocovq99b.fsf@gnu.org>
References: <83ocovq99b.fsf@gnu.org>
Message-ID: <4AC11554.8070406@behdad.org>

Hi Eli,

I'm not sure I understand the issue you are raising completely.  However, I 
think it's a non-issue since BN characters are conceptually removed in X9, so 
it doesn't matter.  However, there is a change around these rules coming in 
the next update to UAX#9 that explicitly supports the code as is, see:

   http://www.unicode.org/reports/tr9/tr9-20.html

Does that address your concern?

behdad

On 09/28/2009 01:17 PM, Eli Zaretskii wrote:
> Hi,
>
> Sorry if this is a FAQ, and please point me at it if it is.
>
> The file fribidi-bidi.c in the FriBidi library has this code:
>
>      for_run_list (pp, main_run_list)
>      {
>        FriBidiCharType this_type = RL_TYPE (pp);
>        if (FRIBIDI_IS_EXPLICIT_OR_BN (this_type))
> 	{
> 	  ...
> 	}
>        else
> 	{
> 	  /* X6. For all types besides RLE, LRE, RLO, LRO, and PDF:
> 	     a. Set the level of the current character to the current
> 	     embedding level.
> 	     b. Whenever the directional override status is not neutral,
> 	     reset the current character type to the directional override
> 	     status. */
> 	  RL_LEVEL (pp) = level;
> 	  if (!FRIBIDI_IS_NEUTRAL (override))
> 	    RL_TYPE (pp) = override;
> 	}
>
> Unless I'm misreading the code, this means that X6 is not applied to
> characters whose original type is "WEAK BN".  However, UAX#9 does not
> exempt WEAK BN characters from the override, so this looks like being
> against the UAX#9 rules.
>
> Am I missing something? are there good reasons for behaving like that?
>
> TIA
> _______________________________________________
> fribidi mailing list
> fribidi at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/fribidi
>

From eliz at gnu.org  Mon Sep 28 13:57:16 2009
From: eliz at gnu.org (Eli Zaretskii)
Date: Mon, 28 Sep 2009 22:57:16 +0200
Subject: [FriBidi] Implementation of X6 in the FriBidi library
In-Reply-To: <4AC11554.8070406@behdad.org>
References: <83ocovq99b.fsf@gnu.org> <4AC11554.8070406@behdad.org>
Message-ID: <83eipqrdn7.fsf@gnu.org>

> Date: Mon, 28 Sep 2009 15:58:12 -0400
> From: Behdad Esfahbod <behdad at behdad.org>
> CC: fribidi at lists.freedesktop.org
> 
> I'm not sure I understand the issue you are raising completely.  However, I 
> think it's a non-issue since BN characters are conceptually removed in X9, so 
> it doesn't matter.

It does matter in an implementation that retains the format codes
instead of removing them, according to section 5.2 of UAX#9.

> However, there is a change around these rules coming in 
> the next update to UAX#9 that explicitly supports the code as is, see:
> 
>    http://www.unicode.org/reports/tr9/tr9-20.html
> 
> Does that address your concern?

Yes, thank you.