Re: [Ltru] RE: ISO 639-2 decision: "mis"

Hi -

> From: "Addison Phillips" <addison@yahoo-inc.com>
> To: "Randy Presuhn" <randy_presuhn@mindspring.com>
> Cc: "LTRU Working Group" <ltru@ietf.org>
> Sent: Friday, June 15, 2007 11:19 AM
> Subject: Re: [Ltru] RE: ISO 639-2 decision: "mis"
>
> Randy Presuhn wrote:
> > 
> > I disagree with the phrase "or when the range of language tags supported in
> > a given application are constrained."  If the application knows what language
> > it is, it should use the correct tag, if one exists.  If the application doesn't
> > know what language is in use, "und" would be correct.
> 
> That's nice in theory. But some applications use a subset of tags (I 
> used the MARC21 example in the text) and then transmit them through 
> another system (where the larger range of RFC 4646 is available). Those 
> systems only know that the content is 'mis' (because it is tagged that 
> way) and not what the miscellaneous language happens to be.

This is an interworking problem, where data from a non-BCP 47 environment
needs to travel through a BCP 47 one.  In this particular case, things
work out.  But as for the proposed text...

...
> The 'mis' (Uncoded) primary language subtag is used to identify 
> linguistic content whose language is known but cannot otherwise be 
> identified.

The phrase "known but cannot otherwise be identified" is a bit obscure.
I suggest replacing it with "known but for which no subtag has been defined."

> It is intended for use when the range of language tags is 
> constrained or for languages not otherwise categorized.

This only makes sense if we somehow want to shoehorn MARC into being
BCP 47 compliant.  I propose deleting this sentence.

> It SHOULD NOT be 
> used except when other means of identifying the language are not 
> available.

Minor edit: I think the would be much easier to read if worded:
 It SHOULD NOT be used when other means of identifying the language are available.

> For example, a library application might be limited to the 
> set of subtags defined for use by the [MARC21] standard. The 'mis' 
> subtag might be used by this application for languages not included in 
> that set.

I think this is not a good example of a BCP 47 use of "mis".

Although I think it's quite reasonable and fortunate that MARC data
could traverse a BCP 47 environment unharmed (and this falls under the
"SHOULD NOT" rules of RFC 2119), I do *not* think we should stretch things
to where we'd claim that MARC is "compliant".  It's just a happy circumstance
that data encoded under the MARC regime is still meaningful when interpreted
under BCP 47 rules, unless tagged with "mis".

Randy

_______________________________________________
Ltru mailing list
Ltru@ietf.org
https://www1.ietf.org/mailman/listinfo/ltru