Re: [Crisp] Last Call Comments on common-transport-03

Remember when 3470 was written.  RelaxNG et al. were still quite new and unproven.  Even XML Schema was pretty new, but the general sentiment was that it was "better" than DTDs.  If I had to write it over again the only thing I'd be inclined to change would be adding a sentence to emphasize the "use what makes sense" point.

xml:lang: We didn't harp on that because tool support for it was mixed at the time.  I personally believe that it should be used today in new specs  instead of creating a new attribute to carry the same info, but I don't see the need to change an older spec if a "custom" attribute was used.  That's what I had to do in EPP,for example.

-Scott- 

 -----Original Message-----
From: 	Andrew Newton [mailto:andy@hxr.us]
Sent:	Wednesday, August 23, 2006 01:37 PM Eastern Standard Time
To:	Hollenbeck, Scott
Cc:	Marcos Sanz/Denic; crisp@ietf.org; iesg@ietf.org
Subject:	Re: [Crisp] Last Call Comments on common-transport-03

Hollenbeck, Scott wrote:
> Andy,
> 
> I don't really think that 3470 has been the cause of confusion. If anything, it has helped to reduce confusion.  The real problem has been that people don't know that it exists, or they haven't really read it.  As you noted, the situation with RelaxNG,  for example,  is typically clarified once people read what the document actually says.
> 
> (Sorry for top-posting.  I'm doing this from my Treo.)
> 
> -Scott-
> (One of the authors of 3470)

Scott,

3470 talks about XML Schema in about 8 separate places, many of which are 
"here is a recommended way of doing X in XML Schema".  With the exception of 
DTDs, all other schema languages are only mentioned in a small paragraph 
that can be paraphrased as saying, "There are other schema languages." 
Given the now vast number of IETF specifications using XML Schema, it is 
easy for one to conclude that XML Schema is the IETF way, even if 
unofficially in the way that re-using TXT records can cause headaches.

But given you are one of the authors of 3470, perhaps you can answer the 
question at hand.  Does 3470 recommend using xml:lang over something like 
<attribute name="lang" type="language"/> in XML Schema, even though the 
content of both is defined by RFC 3066?  Is the qualification of the element 
name necessary in understanding RFC 3066?  Or is the guidance more general, 
in that XML content that is human language be attributed with 3066 language 
tags so that it might be interpreted in language-dependent contexts?

-andy