Why does it take so long?

Harald Tveit Alvestrand harald at alvestrand.no
Fri Jan 9 18:03:10 CET 2004

--On 9. januar 2004 08:12 -0800 Mark Davis <mark.davis at jtcsv.com> wrote:

> 15 is nicer than 16 because 0..15 fits in 4 bits. Not a major issue, but
> since it is such a minor change, and because 16 is chosen to have plenty
> of discrimination space, it's a good change.

but no tags have zero length, so you can start counting from one...

> We tried to be careful to produce a grammar that accepted all generated
> codes from RFC 3066, plus it grandfathered in all registered codes. So it
> would be a complete superset of everything that worked under RFC 3066. It
> would not allow in all codes that *could have been registered*, but that
> is not a problem once it goes into effect and the registry switches over.
> If you think that for any reason we have not accomplished this, please
> let us know.

I'm more worried about the grammar for 3066 not accepting all tags 
generated by the new grammar - this can lead to backwards compatibility 
issues, where parsers built to the old grammar will reject "new" tags.

Likely, this is a non-problem, because most parsers just scan the whole 
thing in text mode anyway, but people who are on the list and know the code 
of those parsers should speak up now.


More information about the Ietf-languages mailing list