[YES] The Linguasphere proposal is suited to RFC 3066 (or itssuccessors) and its consuming protocols

Peter Constable petercon at microsoft.com
Mon Jun 7 17:36:20 CEST 2004

> From: ietf-languages-bounces at alvestrand.no [mailto:ietf-languages-
> bounces at alvestrand.no] On Behalf Of Clay Compton

> What can I say; maybe I just enjoy being contrary.  However, I think
adding *parts*
> of the Linguasphere proposal the RFC 3066 can be beneficial.  For one
thing, it would
> cut back on the number of custom tags requested in this forum, which
most RFC 3066
> implementers don't seem to notice, anyway.

Two issues: 

First, you have said "parts". Thus, an RFC would have to have some basis
to determine which parts were or were not to be permitted.

Secondly, you have said that the number of custom tag requests would be
cut back, but I question how big an issue that would be. Sure, if
someone wanted to request "prsl", they wouldn't have to. But I've
already pointed out ways in which the leaf nodes may not partition
varieties in the way a given user may need. For instance, if I were
looking for fine granularity for N. Thai, I'd still be requesting custom
tags since those being provided do not make distinctions the way I'd
want to use them. 

Also, you've said that implementers don't seem to notice the custom
tags, but do you really think many implementers are going to try to make
sense of 25000 entities? I actually think we'd be better off with 500
custom registrations that we know are being used somewhere than 25000
IDs about which we know nothing: I suggest that the 500 are more likely
to be supported by implementers than the 25000.

> However, "cy-
> cyde-prsl" is a perfectly valid tag in RFC 3066 today, it accurately
reflects that the
> tagged language variety is related to Welsh (which makes it more
> satisfying),

Well, aesthetics are not a particular requirement here, IMO. As you
note, though, the hierarchical coding of rootnode-subnode-subnode can be
useful for query-servicing and fallback purposes: a resource tagged
"cy-cyde-prsl" can appropriately be returned for a request for "cy" or
"cyde", and if a resource tagged "cy-cyde-prsl" is requested but not
available, then "cy-cyde" and "cy" are likely to be an acceptable
fallback chain (though what implementations really need for resource
fallback chains are in general *not necessarily* that simple). 

But, as Addison has already pointed out, "cy-cyde-prsl" would be
incompatible with the proposed RFC wrt aspects that we're not going to
change at this point.

So, I have no problem with the hierarchical coding. The use of alpha-4
is a problem, so let's suppose that Linguasphere provided us with (say)
alpha-5 alternatives (or, say, we simply concatenate "l" & alpha-4): I'd
have no problem if someone wanted to use a tag of that form. But, I *do*
have concerns with proposing that all such Linguasphere tags are
supported in an RFC, which concerns I have already voiced.

> Of course, I'd
> like to hear the Linguasphere folks pledge that they'll avoid any tag
name collisions
> with ISO 15924.

This is *not* a good idea, IMO.

Peter Constable
Globalization Infrastructure and Font Technologies
Microsoft Windows Division

More information about the Ietf-languages mailing list