25 Jan 18:03
Extension Relation Type Comparison - LC Comment on draft-nottingham-http-plink-header-07
Eran Hammer-Lahav <eran <at> hueniverse.com>
2010-01-25 17:03:17 GMT
2010-01-25 17:03:17 GMT
The current text about comparing extension relation types is unclear: When extension relation types are compared, they MUST be compared as URIs in a case-insensitive fashion, character-by-character. Because of this, all-lowercase URIs SHOULD be used for extension relations. What does it mean "compared as URIs"? It is clear that these two URIs would be deemed equivalent: http://example.com/rel/type HTTP://example.COM/rel/TYPE But are they also equivalent to: http://example.com:80/rel/type I would argue that normalizing URIs in this context is non-obvious and counter-intuitive. We are not using URIs for the purpose of resolving them, but only as a way to construct unique strings, using the URI format to define authority and minting rules. I think we would be better off comparing extension relation types "as strings in a case-insensitive fashion, character-by-character". EHL
RSS Feed