Authoring HTML: Handling Right-to-left Scripts, additional changes

[Richard Ishida]

I have implemented the feedback received on this document in the editor's version at
http://www.w3.org/International/docs/bp-html-bidi/

I have also, after further discussions, made some further changes as follows:

1. changed Best Practices to Techniques throughout (and just removed from the title)

2. created 3 new techniques
http://www.w3.org/International/docs/bp-html-bidi/#scrollbar
http://www.w3.org/International/docs/bp-html-bidi/#iso_encoding
http://www.w3.org/International/docs/bp-html-bidi/#externalized_text

3. redesigned http://www.w3.org/International/docs/bp-html-bidi/uanotes and changed the links after each technique so that there is only one now, pointing to a section in the uanotes document.  The uanotes document contains a short summary of any browser-specific note and pointers to tests and results for more information.  It also contains links to the appropriate section or sections of the techniques index so you can find more information. This document still needs some more work.


If you have any comments on the above, please send to this list.  Barring showstoppers, I'd like to publish the final version in about a week to ten days.

Thanks.
RI


============
Richard Ishida
Internationalization Lead
W3C (World Wide Web Consortium)

http://www.w3.org/International/
http://rishida.net/

[Matitiahu Allouche]

I have some comments about the ISO Encoding HowTo (at http://www.w3.org/International/docs/bp-html-bidi/#iso_encoding )

1) We find the sentence:
'Visual' refers to the practice of typing in the Hebrew characters in reverse order and preventing automatic line breaks.

This is not a good definition of 'Visual', since there are input methods which allow to type Hebrew in natural order while storing it in visual order. Also, preventing automatic line breaks is recommended when the text is in visual order (to avoid text corruption when text is reflowed automatically), but it is in no way an essential attribute of visual text.  I suggest to stick to a more classic definition of 'Visual', such as:
'Visual' refers to the practice of storing Hebrew characters in presentation order, so that there is no reliance on reordering performed by the operating system or the display subsystem.

2) "all characters in memory in the order" => "all characters are stored in memory in the order"

3) We find the sentence:
Because HTML uses the Unicode bidirectional algorithm, conforming documents encoded using ISO 8859-8 must be labeled as ISO-8859-8-i.

This sentence is problematic, IMHO: HTML is a protocol and does not use the Unicode bidi algorithm.  HTML assumes that Bidi data are stored in logical order (with the exception of charset="ISO-8859-8"), thus HTML rendering agents must use the Unicode bidi algorithm to present the bidi data in correct visual order.  I suggest the following phrasing:
HTML assumes by default that Bidi data is stored in logical order, to that rendering agents will have to use the Unicode bidirectional algorithm to present the text in correct visual order.  If the encoding is ISO-8859-9, the corresponding charset specification must be ISO-8859-8-i.

4) "ISO-8859-6 (Arabic) is not visual ordering" => "ISO-8859-6 (Arabic) does not imply visual ordering"

Shalom (Regards),  Mati
          Bidi Architect
          Globalization Center Of Competency - Bidirectional Scripts
          IBM Israel
          Phone: [phone]    Fax: [phone]    Mobile: [phone]

From:	"Richard Ishida" <ishida@...>
To:	<www-international@...>
Date:	13/08/2009 22:10
Subject:	Authoring HTML: Handling Right-to-left Scripts, additional changes
Sent by:	www-international-request@...

---

I have implemented the feedback received on this document in the editor's version at
http://www.w3.org/International/docs/bp-html-bidi/

I have also, after further discussions, made some further changes as follows:

1. changed Best Practices to Techniques throughout (and just removed from the title)

2. created 3 new techniques
http://www.w3.org/International/docs/bp-html-bidi/#scrollbar
http://www.w3.org/International/docs/bp-html-bidi/#iso_encoding
http://www.w3.org/International/docs/bp-html-bidi/#externalized_text

3. redesigned http://www.w3.org/International/docs/bp-html-bidi/uanotes and changed the links after each technique so that there is only one now, pointing to a section in the uanotes document.  The uanotes document contains a short summary of any browser-specific note and pointers to tests and results for more information.  It also contains links to the appropriate section or sections of the techniques index so you can find more information. This document still needs some more work.


If you have any comments on the above, please send to this list.  Barring showstoppers, I'd like to publish the final version in about a week to ten days.

Thanks.
RI


============
Richard Ishida
Internationalization Lead
W3C (World Wide Web Consortium)

http://www.w3.org/International/
http://rishida.net/

[Matitiahu Allouche]

This is a comment on the externalized text howto (at http://www.w3.org/International/docs/bp-html-bidi/#externalized_text ).

As far as I remember, putting LRE/RLE and PDF around a Javascript message gives the required effect in Internet Explorer only if the message is displayed on one line.  If the message spans multiple lines, only the first line is affected.  At least this is what happens for me with MS IE7 under WinXP SP2.  I have not checked what happens with other browsers and other platforms.

This bug/feature does not apply to titles and tooltips which are always displayed as a single line.


Shalom (Regards),  Mati
          Bidi Architect
          Globalization Center Of Competency - Bidirectional Scripts
          IBM Israel
          Phone: [phone]    Fax: [phone]    Mobile: [phone]

From:	"Richard Ishida" <ishida@...>
To:	<www-international@...>
Date:	13/08/2009 22:10
Subject:	Authoring HTML: Handling Right-to-left Scripts, additional changes
Sent by:	www-international-request@...

---

I have implemented the feedback received on this document in the editor's version at
http://www.w3.org/International/docs/bp-html-bidi/

I have also, after further discussions, made some further changes as follows:

1. changed Best Practices to Techniques throughout (and just removed from the title)

2. created 3 new techniques
http://www.w3.org/International/docs/bp-html-bidi/#scrollbar
http://www.w3.org/International/docs/bp-html-bidi/#iso_encoding
http://www.w3.org/International/docs/bp-html-bidi/#externalized_text

3. redesigned http://www.w3.org/International/docs/bp-html-bidi/uanotes and changed the links after each technique so that there is only one now, pointing to a section in the uanotes document.  The uanotes document contains a short summary of any browser-specific note and pointers to tests and results for more information.  It also contains links to the appropriate section or sections of the techniques index so you can find more information. This document still needs some more work.


If you have any comments on the above, please send to this list.  Barring showstoppers, I'd like to publish the final version in about a week to ten days.

Thanks.
RI


============
Richard Ishida
Internationalization Lead
W3C (World Wide Web Consortium)

http://www.w3.org/International/
http://rishida.net/

[John Cowan]

Matitiahu Allouche scripsit:

>  I suggest to stick to a more classic definition of 'Visual', such as:
> 'Visual' refers to the practice of storing Hebrew characters in
> presentation order, so that there is no reliance on reordering performed
> by the operating system or the display subsystem.

Except, what is "presentation order"?  It can be rightward, leftward,
or upward nowadays.  To speak of left to right order as "presentation
order" is to submit to a purely parochial viewpoint.

I would go with "the practice of storing Hebrew characters in
left-to-right order on each line of text", combined with an explanation
that the term "visual" is a historical one rooted in the days when only
left-to-right reading was considered normal for computer text.

> This sentence is problematic, IMHO: HTML is a protocol and does not use
> the Unicode bidi algorithm.

HTML is not a protocol but a format.

> I suggest the following phrasing:
> HTML assumes by default that Bidi data is stored in logical order, to that
> present the text in correct visual order.  If the encoding is ISO-8859-9,
> the corresponding charset specification must be ISO-8859-8-i.

(The mention of 8859-9 is clearly a typo here.)

Again, this is a biased wording.  I would go with:

The charset specifies whether a document encoded in 8859-8 is in visual or
logical order.  Use the charset "iso-8859-8-i" for logical order encoding
(preferred), or "iso-8859-8" for visual-order encoding (deprecated).
Text in Unicode and other encodings is always in logical order.

--
We call nothing profound                        cowan@...
that is not wittily expressed.                  John Cowan
        --Northrop Frye (improved)

["Martin J. Dürst"]

On 2009/08/26 11:03, John Cowan wrote:

> Matitiahu Allouche scripsit:
>
>>   I suggest to stick to a more classic definition of 'Visual', such as:
>> 'Visual' refers to the practice of storing Hebrew characters in
>> presentation order, so that there is no reliance on reordering performed
>> by the operating system or the display subsystem.
>
> Except, what is "presentation order"?  It can be rightward, leftward,
> or upward nowadays.  To speak of left to right order as "presentation
> order" is to submit to a purely parochial viewpoint.
>
> I would go with "the practice of storing Hebrew characters in
> left-to-right order on each line of text", combined with an explanation
> that the term "visual" is a historical one rooted in the days when only
> left-to-right reading was considered normal for computer text.

I agree with your comments re. presentation order. But the word 'visual'
is more than historical, and is justified by more than a simple
'parochial' LTR assumption. Visual contains explicit linebreaks, and
does not use subruns that go the other way (for 'logical', such subruns
are used at least for numbers, if not for embeddings). An immaginary
"RTL visual" would have the same properties, and (except for special
lucky cases) the same disadvantages (a pain to input, edit, vocalize,...).

Regards,   Martin.

--
#-# Martin J. Dürst, Professor, Aoyama Gakuin University
#-# http://www.sw.it.aoyama.ac.jp   mailto:duerst@...

[John Cowan]

"Martin J. Dürst" scripsit:

> I agree with your comments re. presentation order. But the word 'visual'
> is more than historical, and is justified by more than a simple
> 'parochial' LTR assumption. Visual contains explicit linebreaks, and
> does not use subruns that go the other way (for 'logical', such subruns
> are used at least for numbers, if not for embeddings).

Fair enough.

--
Unless it was by accident that I had            John Cowan
offended someone, I never apologized.           cowan@...
        --Quentin Crisp                         http://www.ccil.org/~cowan

[Najib Tounsi-2]

Matitiahu Allouche wrote:

>
> This is a comment on the externalized text howto (at
> http://www.w3.org/International/docs/bp-html-bidi/#externalized_text ).
>
> As far as I remember, putting LRE/RLE and PDF around a Javascript
> message gives the required effect in Internet Explorer only if the
> message is displayed on one line.  If the message spans multiple
> lines, only the first line is affected.  At least this is what happens
> for me with MS IE7 under WinXP SP2.  I have not checked what happens
> with other browsers and other platforms.

This looks fine (putting RLE and PDF around a Javascript message which
spans multiple lines)
 with other browsers on MacOSX 10.5.7 : Safari 3.2, Opera 9.5, Firefox
3, Mozilla 1.1.16

Salam, Najib

>
> This bug/feature does not apply to titles and tooltips which are
> always displayed as a single line.
>
>
> Shalom (Regards),  Mati
>           Bidi Architect
>           Globalization Center Of Competency - Bidirectional Scripts
>           IBM Israel
>           Phone: [phone]    Fax: [phone]    Mobile: +972
> 52 2554160
>
>
>
> From: "Richard Ishida" <ishida@...>
> To: <www-international@...>
> Date: 13/08/2009 22:10
> Subject: Authoring HTML: Handling Right-to-left Scripts, additional
> changes
> Sent by: www-international-request@...
>
>
> ------------------------------------------------------------------------
>
>
>
> I have implemented the feedback received on this document in the
> editor's version at
> http://www.w3.org/International/docs/bp-html-bidi/
>
> I have also, after further discussions, made some further changes as
> follows:
>
> 1. changed Best Practices to Techniques throughout (and just removed
> from the title)
>
> 2. created 3 new techniques
> http://www.w3.org/International/docs/bp-html-bidi/#scrollbar
> http://www.w3.org/International/docs/bp-html-bidi/#iso_encoding
> http://www.w3.org/International/docs/bp-html-bidi/#externalized_text
>
> 3. redesigned
> http://www.w3.org/International/docs/bp-html-bidi/uanotes and changed
> the links after each technique so that there is only one now, pointing
> to a section in the uanotes document.  The uanotes document contains a
> short summary of any browser-specific note and pointers to tests and
> results for more information.  It also contains links to the
> appropriate section or sections of the techniques index so you can
> find more information. This document still needs some more work.
>
>
> If you have any comments on the above, please send to this list.
>  Barring showstoppers, I'd like to publish the final version in about
> a week to ten days.
>
> Thanks.
> RI
>
>
> ============
> Richard Ishida
> Internationalization Lead
> W3C (World Wide Web Consortium)
>
> http://www.w3.org/International/
> http://rishida.net/
>

--
Najib TOUNSI (mailto:tounsi @ w3.org)
W3C Office in Morocco (http://www.w3c.org.ma/)
Ecole Mohammadia d'Ingénieurs, BP. 765 Agdal-RABAT Morocco
Phone : [phone] (P1711)  Fax : [phone]
Mobile: [phone]

[Richard Ishida]

Some parts of this message have been removed. Learn more about Nabble's security policy.

Hmm.  I went to some lengths to test that on IE7, IE8 and some other browsers, but couldn't replicate it.  I used a modified version of tests 20-22 at http://www.w3.org/International/tests/tests-html-css/tests-bidi-chrome/generate?test=20

 

If we do find repeatable evidence of the problem, we can add the information to http://www.w3.org/International/docs/bp-html-bidi/uanotes.html

 

Cheers,

RI

 

============
Richard Ishida
Internationalization Lead
W3C (World Wide Web Consortium)

http://www.w3.org/International/
http://rishida.net/

 

From: Matitiahu Allouche [mailto:matial@...]
Sent: 26 August 2009 01:57
To: Richard Ishida
Cc: www-international@...
Subject: Re: Authoring HTML: Handling Right-to-left Scripts, additional changes

 

This is a comment on the externalized text howto (at http://www.w3.org/International/docs/bp-html-bidi/#externalized_text ).

As far as I remember, putting LRE/RLE and PDF around a Javascript message gives the required effect in Internet Explorer only if the message is displayed on one line.  If the message spans multiple lines, only the first line is affected.  At least this is what happens for me with MS IE7 under WinXP SP2.  I have not checked what happens with other browsers and other platforms.

This bug/feature does not apply to titles and tooltips which are always displayed as a single line.


Shalom (Regards),  Mati
          Bidi Architect
          Globalization Center Of Competency - Bidirectional Scripts
          IBM Israel
          Phone: [phone]    Fax: [phone]    Mobile: [phone]

From:	"Richard Ishida" <ishida@...>
To:	<www-international@...>
Date:	13/08/2009 22:10
Subject:	Authoring HTML: Handling Right-to-left Scripts, additional changes
Sent by:	www-international-request@...

 

---

I have implemented the feedback received on this document in the editor's version at
http://www.w3.org/International/docs/bp-html-bidi/

I have also, after further discussions, made some further changes as follows:

1. changed Best Practices to Techniques throughout (and just removed from the title)

2. created 3 new techniques
http://www.w3.org/International/docs/bp-html-bidi/#scrollbar
http://www.w3.org/International/docs/bp-html-bidi/#iso_encoding
http://www.w3.org/International/docs/bp-html-bidi/#externalized_text

3. redesigned http://www.w3.org/International/docs/bp-html-bidi/uanotes and changed the links after each technique so that there is only one now, pointing to a section in the uanotes document.  The uanotes document contains a short summary of any browser-specific note and pointers to tests and results for more information.  It also contains links to the appropriate section or sections of the techniques index so you can find more information. This document still needs some more work.


If you have any comments on the above, please send to this list.  Barring showstoppers, I'd like to publish the final version in about a week to ten days.

Thanks.
RI


============
Richard Ishida
Internationalization Lead
W3C (World Wide Web Consortium)

http://www.w3.org/International/
http://rishida.net/

[Richard Ishida]

Some parts of this message have been removed. Learn more about Nabble's security policy.

Changes made. Thanks!

 

RI

 

============
Richard Ishida
Internationalization Lead
W3C (World Wide Web Consortium)

http://www.w3.org/International/
http://rishida.net/

 

From: Matitiahu Allouche [mailto:matial@...]
Sent: 26 August 2009 01:57
To: Richard Ishida
Cc: www-international@...
Subject: Re: Authoring HTML: Handling Right-to-left Scripts, additional changes

 

I have some comments about the ISO Encoding HowTo (at http://www.w3.org/International/docs/bp-html-bidi/#iso_encoding )

1) We find the sentence:
'Visual' refers to the practice of typing in the Hebrew characters in reverse order and preventing automatic line breaks.

This is not a good definition of 'Visual', since there are input methods which allow to type Hebrew in natural order while storing it in visual order. Also, preventing automatic line breaks is recommended when the text is in visual order (to avoid text corruption when text is reflowed automatically), but it is in no way an essential attribute of visual text.  I suggest to stick to a more classic definition of 'Visual', such as:
'Visual' refers to the practice of storing Hebrew characters in presentation order, so that there is no reliance on reordering performed by the operating system or the display subsystem.

2) "all characters in memory in the order" => "all characters are stored in memory in the order"

3) We find the sentence:
Because HTML uses the Unicode bidirectional algorithm, conforming documents encoded using ISO 8859-8 must be labeled as ISO-8859-8-i.

This sentence is problematic, IMHO: HTML is a protocol and does not use the Unicode bidi algorithm.  HTML assumes that Bidi data are stored in logical order (with the exception of charset="ISO-8859-8"), thus HTML rendering agents must use the Unicode bidi algorithm to present the bidi data in correct visual order.  I suggest the following phrasing:
HTML assumes by default that Bidi data is stored in logical order, to that rendering agents will have to use the Unicode bidirectional algorithm to present the text in correct visual order.  If the encoding is ISO-8859-9, the corresponding charset specification must be ISO-8859-8-i.

4) "ISO-8859-6 (Arabic) is not visual ordering" => "ISO-8859-6 (Arabic) does not imply visual ordering"

Shalom (Regards),  Mati
          Bidi Architect
          Globalization Center Of Competency - Bidirectional Scripts
          IBM Israel
          Phone: [phone]    Fax: [phone]    Mobile: [phone]

From:	"Richard Ishida" <ishida@...>
To:	<www-international@...>
Date:	13/08/2009 22:10
Subject:	Authoring HTML: Handling Right-to-left Scripts, additional changes
Sent by:	www-international-request@...

 

---

I have implemented the feedback received on this document in the editor's version at
http://www.w3.org/International/docs/bp-html-bidi/

I have also, after further discussions, made some further changes as follows:

1. changed Best Practices to Techniques throughout (and just removed from the title)

2. created 3 new techniques
http://www.w3.org/International/docs/bp-html-bidi/#scrollbar
http://www.w3.org/International/docs/bp-html-bidi/#iso_encoding
http://www.w3.org/International/docs/bp-html-bidi/#externalized_text

3. redesigned http://www.w3.org/International/docs/bp-html-bidi/uanotes and changed the links after each technique so that there is only one now, pointing to a section in the uanotes document.  The uanotes document contains a short summary of any browser-specific note and pointers to tests and results for more information.  It also contains links to the appropriate section or sections of the techniques index so you can find more information. This document still needs some more work.


If you have any comments on the above, please send to this list.  Barring showstoppers, I'd like to publish the final version in about a week to ten days.

Thanks.
RI


============
Richard Ishida
Internationalization Lead
W3C (World Wide Web Consortium)

http://www.w3.org/International/
http://rishida.net/

[CE Whitehead]

Some parts of this message have been removed. Learn more about Nabble's security policy.

Hi.

> Date: Tue, 25 Aug 2009 22:03:49 -0400
> To: matial@...
> CC: ishida@...; www-international@...
> From: cowan@...
> Subject: Re: Authoring HTML: Handling Right-to-left Scripts, additional changes
>
> Matitiahu Allouche scripsit:
>
> > I suggest to stick to a more classic definition of 'Visual', such as:
> > 'Visual' refers to the practice of storing Hebrew characters in
> > presentation order, so that there is no reliance on reordering performed
> > by the operating system or the display subsystem.
>
> Except, what is "presentation order"? It can be rightward, leftward,
> or upward nowadays. To speak of left to right order as "presentation
> order" is to submit to a purely parochial viewpoint.

 
Hmm, o.k., but I think Mati's suggestion is fine here--given the topic of Richard Ishida's text.

>
> I would go with "the practice of storing Hebrew characters in
> left-to-right order on each line of text", combined with an explanation
> that the term "visual" is a historical one rooted in the days when only
> left-to-right reading was considered normal for computer text.
>
> > This sentence is problematic, IMHO: HTML is a protocol and does not use
> > the Unicode bidi algorithm.
>
> HTML is not a protocol but a format.

Right; HTTP or HTTPS (or FTP) is the protocol, as I understand it.  HTML is a markup language.  I did not have a problem with Richard Ishida's sentence here, however--maybe i need enlightening.  HTML and XML pages allow text to be displayed and the text can be stored in various character encodings including unicode; I don't know if this is using the unicode bidi algorithm or not. 

>
> > I suggest the following phrasing:
> > HTML assumes by default that Bidi data is stored in logical order, to that
> > present the text in correct visual order. If the encoding is ISO-8859-9,
> > the corresponding charset specification must be ISO-8859-8-i.
>
> (The mention of 8859-9 is clearly a typo here.)
>
> Again, this is a biased wording. I would go with:
>
> The charset specifies whether a document encoded in 8859-8 is in visual or
> logical order. Use the charset "iso-8859-8-i" for logical order encoding
> (preferred), or "iso-8859-8" for visual-order encoding (deprecated).
> Text in Unicode and other encodings is always in logical order.
>
> --
> We call nothing profound cowan@...
> that is not wittily expressed. John Cowan
> --Northrop Frye (improved)
>

Best,
 
C. E. Whitehead
cewcathar@...

[Matitiahu Allouche]

Some parts of this message have been removed. Learn more about Nabble's security policy.

Here is my test file:



Shalom (Regards),  Mati
          Bidi Architect
          Globalization Center Of Competency - Bidirectional Scripts
          IBM Israel
          Phone: [phone]    Fax: [phone]    Mobile: [phone]

From:	"Richard Ishida" <ishida@...>
To:	Matitiahu Allouche/Israel/IBM@IBMIL
Cc:	<www-international@...>
Date:	26/08/2009 18:54
Subject:	RE: Authoring HTML: Handling Right-to-left Scripts, additional changes

---

Hmm.  I went to some lengths to test that on IE7, IE8 and some other browsers, but couldn't replicate it.  I used a modified version of tests 20-22 at http://www.w3.org/International/tests/tests-html-css/tests-bidi-chrome/generate?test=20
 
If we do find repeatable evidence of the problem, we can add the information to http://www.w3.org/International/docs/bp-html-bidi/uanotes.html
 
Cheers,
RI
 
============
Richard Ishida
Internationalization Lead
W3C (World Wide Web Consortium)

http://www.w3.org/International/
http://rishida.net/


 
From: Matitiahu Allouche [mailto:matial@...]
Sent: 26 August 2009 01:57
To: Richard Ishida
Cc: www-international@...
Subject: Re: Authoring HTML: Handling Right-to-left Scripts, additional changes
 

This is a comment on the externalized text howto (at http://www.w3.org/International/docs/bp-html-bidi/#externalized_text ).

As far as I remember, putting LRE/RLE and PDF around a Javascript message gives the required effect in Internet Explorer only if the message is displayed on one line.  If the message spans multiple lines, only the first line is affected.  At least this is what happens for me with MS IE7 under WinXP SP2.  I have not checked what happens with other browsers and other platforms.

This bug/feature does not apply to titles and tooltips which are always displayed as a single line.


Shalom (Regards),  Mati
         Bidi Architect
         Globalization Center Of Competency - Bidirectional Scripts
         IBM Israel
         Phone: [phone]    Fax: [phone]    Mobile: [phone]

From:	"Richard Ishida" <ishida@...>
To:	<www-international@...>
Date:	13/08/2009 22:10
Subject:	Authoring HTML: Handling Right-to-left Scripts, additional changes
Sent by:	www-international-request@...

 

---

I have implemented the feedback received on this document in the editor's version at
http://www.w3.org/International/docs/bp-html-bidi/

I have also, after further discussions, made some further changes as follows:

1. changed Best Practices to Techniques throughout (and just removed from the title)

2. created 3 new techniques
http://www.w3.org/International/docs/bp-html-bidi/#scrollbar
http://www.w3.org/International/docs/bp-html-bidi/#iso_encoding
http://www.w3.org/International/docs/bp-html-bidi/#externalized_text

3. redesigned http://www.w3.org/International/docs/bp-html-bidi/uanotes and changed the links after each technique so that there is only one now, pointing to a section in the uanotes document.  The uanotes document contains a short summary of any browser-specific note and pointers to tests and results for more information.  It also contains links to the appropriate section or sections of the techniques index so you can find more information. This document still needs some more work.


If you have any comments on the above, please send to this list.  Barring showstoppers, I'd like to publish the final version in about a week to ten days.

Thanks.
RI


============
Richard Ishida
Internationalization Lead
W3C (World Wide Web Consortium)

http://www.w3.org/International/
http://rishida.net/

First message does not have RLE/PDF.

Second message does have RLE/PDF.