Re: Date format glitch

On 10.10.2010 23:08, Daniel Stenberg wrote:
> Hi friends,
>
> In RFC2616 section 3.3.1 it says that "HTTP applications have
> historically allowed three different formats" and one of them is the
> RFC822/RFC1123 format.
>
> It then goes on to say that RFC822 was updated by RFC1123 and that
> "clients and servers that parse the date value MUST accept all three
> formats". RFC1123 basically only changed the year field to be 4-digit
> instead of 2-digit.
>
> However - and here's my question/thoughts - the follwing BNF syntax
> description of the rfc1123 time in RFC2616 is incorrect by my reading of
> it. RFC822 has the seconds part of the time triplet (HH:MM:SS) optional.
> RFC822
> defines it as:
>
> hour = 2DIGIT ":" 2DIGIT [":" 2DIGIT]
>
> This "flaw" is in httpbis as well and while I don't think it ever causes
> any particular problem anywhere, it just popped to my attention as I
> believed my parser that parses HTTP dates were RFC822 compliant when in
> fact it wasn't... (It was used to parse RFC822 date strings in a totally
> unrelated context when this was reported.)
>
> It might be worth to note this. Or not. I just wanted to mention this.

Ah, spec archeology :-)

The thing called "rfc1123-date" isn't really the format defined by RFC 
1123. See:

"However, the preferred format is a fixed-length subset of that defined 
by [RFC1123]:" -- 
<http://greenbytes.de/tech/webdav/draft-ietf-httpbis-p1-messaging-11.html#rfc.section.6.1>

Do we need to expand this sentence?

Best regards, Julian

PS: I think we discussed renaming the format some time ago, but 
concluded it would just increase the confusion we already see.

Received on Monday, 11 October 2010 12:43:10 UTC