RE: Issue 163, was: Meaning of invalid but well-formed dates

Daniel Stenberg wrote:
> It does beg the question: are implementations doing the comparisons
> case sensitively when parsing data in HTTP headers? I know libcurl
> certainly doesn't...
> 
> And yeah, I wasn't aware of them being case sensitive either.

The specification does say:

      Recipients of date values are encouraged to be robust in
      accepting date values that may have been sent by non-HTTP
      applications, as is sometimes the case when retrieving or posting
      messages via proxies/gateways to SMTP or NNTP.

SMTP and NNTP dates are not case sensitive, so it makes sense to parse these
in a case-insensitive manner.

But, what does "are encouraged to" mean? Either it should be "are
RECOMMENDED to" (which means "SHOULD") or the statement should go. The
specification should not use non-RFC2119 language when making
recommendations.

And, if it is to be "SHOULD", then the grammar should change to allow the
other cases (probably by adding an obs-rfc5322-date alternative that
references RFC 5322's date). Elsewhere in the document, the grammar reflects
what parsers SHOULD accept, and the prose further restricts what
implementations may generate.

- Brian