Proposal to address I18N-ISSUE-193: reference obs-language-tag instead of defining your own

Proposal to address I18N-ISSUE-193: reference obs-language-tag instead of defining your own
===============================================================

Issue: Section 6.5 (Grammar) defines LANGTAG far more permissively than BCP 47 does--even in its obsolete forms.

(implicit) proposals from I18N:
  1 Provide no LANGTAG production, instead reference the Language-Tag or obs-language-tag production.
    -- hard to read. grammar languages are different. current grammar is copy-and-paste-able into a grammar tool.

  2 Incorporate the Language-Tag production into Turtle's grammar.
    -- The full langtag grammar is bigger than the entire Turtle grammar.

  3 Incorporate the obs-language-tag production into Turtle's grammar.
    -- There are some specs which use the XML grammar notation http://www.w3.org/TR/REC-xml/#sec-notation . This notation does not include {n} or {m,n}. The cost to readers of rolling our own grammar is relatively high; the number of erroneous language tags which are caught by changing from
      1 LANGTAG ::= '@' [a-zA-Z]+ ('-' [a-zA-Z0-9]+)*
to
      2 LANGTAG ::= '@' [a-zA-Z]{1,8} ('-' [a-zA-Z0-9]+){1,8}
is quite low. 2 doesn't catch human errors like "@english" and 1 already catches likly serialized unitialized strings.

  4 Point out that RDF references BCP 47 without complicating the Turtle grammar.
    ++ adopted


Proposal: preserve the current grammar production, adding this text as a comment [[
  Note that the <a href="http://dvcs.w3.org/hg/rdf/raw-file/default/rdf-concepts/index.html#dfn-language-tag">RDF abstract syntax</a> asserts that "the language tag must be well-formed according to section 2.2.9 of [BCP47]".
]]

Please indicate whether this address the stated issue.
-- 
-ericP

Received on Wednesday, 17 October 2012 16:05:32 UTC