RE: New Working Group Note: Best Practices for XML Internationalization - Corrected

Jony, I agree with you.
Of course, when discussing comparison algorithms it depends on the intent, which is why the collation algorithm provides for STRENGTH.
The directional controls can change the semantics so they can't be completely ignored.
 
 


________________________________

	From: www-international-request@w3.org [mailto:www-international-request@w3.org] On Behalf Of rosennej
	Sent: Sunday, February 17, 2008 10:57 AM
	To: 'WWW International'
	Subject: RE: New Working Group Note: Best Practices for XML Internationalization - Corrected
	
	

	I have made some corrections to my previous mail:

	 

	If the string is plain RTL or plain LTR, and the first or last character is not a neutral, the is no need for bidi controls. Otherwise, wrap the string either with an LRM/RLM or with an LRE/RLE and PDF.

	 

	The comparison algorithm should ignore superfluous bidi controls. Some people think the comparison algorithm should ignore all bidi controls, but I am not certain this will produce satisfactory results.

	 

	Jony

	 

	-----Original Message-----

	From: www-international-request@w3.org [mailto:www-international-request@w3.org] On Behalf Of Tex Texin

	Sent: Friday, February 15, 2008 10:27 PM

	To: Douglas Davidson

	Cc: Richard Ishida; WWW International

	Subject: RE: New Working Group Note: Best Practices for XML Internationalization

	 

	 

	You decide on the standardized representation and this can include the presumed context.

	Then you know whether the text needs adjustment when it is placed in a particular environment. 

	 

	-----Original Message-----

	From: Douglas Davidson [mailto:ddavidso@apple.com] 

	Sent: Friday, February 15, 2008 12:19 PM

	To: Tex Texin

	Cc: Richard Ishida; WWW International

	Subject: Re: New Working Group Note: Best Practices for XML Internationalization

	 

	 

	On Feb 15, 2008, at 11:50 AM, Tex Texin wrote:

	 

	> I should be able to extract a string from a database and have it work 

	> equally well regardless of the context.

	 

	The issue of extracting a string from a database and embedding it into a document seems to me to be a particularly knotty one.  How do you form the strings in your database so that they will display well if you do not know in advance whether the context in which they will be embedded will be LTR or RTL?  On the other side, how do you embed strings in your document if you do not know in advance what sort of bidi content they will have?  A mismatched bidi formatting code in an embedded string, for example, could cause problems.

	 

	Douglas Davidson