Re: Alternative to the date regex
+1 on dropping the regex. It isn't from any of the other specs, it isn't specifically called out as explanatory and non-normative, and it is too long to be clear. Some examples would be nice, along with some examples of things which do not conform. wunder --On March 25, 2005 5:11:09 PM + Graham <[EMAIL PROTECTED]> wrote: > > Currently we have this > > "A Date construct is an element whose content MUST conform to the > date-time BNF rule in [RFC3339]. I.e., the content of this element > matches this regular expression: > > [0-9]{8}T[0-9]{2}:[0-9]{2}:[0-9]{2}(\.[0-9]+) > ?(Z|[\+\-][0-9]{2}:[0-9]{2}) > > As a result, the date values conform to the following specifications..." > > The problem with the regex is that it's entirely redundant. If we look at > Norm's message where the regex was suggested [1], he intends it as a profile > of xsd:dateTime, which allows a variety of date formats. However we're using > it as a profile of RFC3339, which already requires that date-times match the > regex 100%. Having the regex there as well is just confusing - until > preparing this email I was under the impression it made some additional > restrictions on RFC3339. > > The nearest thing I see to an additional restriction is that there must be a > capital T between the date and time, which the date-time BNF rule we mention > also requires, but the prose later mentions you might be allowed to use > something different. > > Proposal: > Replace the first para and regex with: > > A Date construct is an element whose content MUST conform to the > date-time BNF rule in [RFC3339]. Note this requires an uppercase letter T > between the date and time sections. > > Secondly, *all* RFC3339 date-times are compatible with the 4 specs mentioned, > so the wording of the second paragraph ("As a result...") is a bit strange, > since it's not as a result of anything we've done. Just say "Date values > expressed in this way are also compatible with...". > > Graham > > [1]http://www.imc.org/atom-syntax/mail-archive/msg13116.html > > -- Walter Underwood Principal Architect, Verity
Re: Alternative to the date regex
Sounds like a plan to me. +1. Robert Sayre Graham wrote: Currently we have this "A Date construct is an element whose content MUST conform to the date-time BNF rule in [RFC3339]. I.e., the content of this element matches this regular expression: [0-9]{8}T[0-9]{2}:[0-9]{2}:[0-9]{2}(\.[0-9]+) ?(Z|[\+\-][0-9]{2}:[0-9]{2}) As a result, the date values conform to the following specifications..." The problem with the regex is that it's entirely redundant. If we look at Norm's message where the regex was suggested [1], he intends it as a profile of xsd:dateTime, which allows a variety of date formats. However we're using it as a profile of RFC3339, which already requires that date-times match the regex 100%. Having the regex there as well is just confusing - until preparing this email I was under the impression it made some additional restrictions on RFC3339. The nearest thing I see to an additional restriction is that there must be a capital T between the date and time, which the date-time BNF rule we mention also requires, but the prose later mentions you might be allowed to use something different. Proposal: Replace the first para and regex with: A Date construct is an element whose content MUST conform to the date-time BNF rule in [RFC3339]. Note this requires an uppercase letter T between the date and time sections. Secondly, *all* RFC3339 date-times are compatible with the 4 specs mentioned, so the wording of the second paragraph ("As a result...") is a bit strange, since it's not as a result of anything we've done. Just say "Date values expressed in this way are also compatible with...". Graham [1]http://www.imc.org/atom-syntax/mail-archive/msg13116.html
Alternative to the date regex
Currently we have this "A Date construct is an element whose content MUST conform to the date-time BNF rule in [RFC3339]. I.e., the content of this element matches this regular expression: [0-9]{8}T[0-9]{2}:[0-9]{2}:[0-9]{2}(\.[0-9]+) ?(Z|[\+\-][0-9]{2}:[0-9]{2}) As a result, the date values conform to the following specifications..." The problem with the regex is that it's entirely redundant. If we look at Norm's message where the regex was suggested [1], he intends it as a profile of xsd:dateTime, which allows a variety of date formats. However we're using it as a profile of RFC3339, which already requires that date-times match the regex 100%. Having the regex there as well is just confusing - until preparing this email I was under the impression it made some additional restrictions on RFC3339. The nearest thing I see to an additional restriction is that there must be a capital T between the date and time, which the date-time BNF rule we mention also requires, but the prose later mentions you might be allowed to use something different. Proposal: Replace the first para and regex with: A Date construct is an element whose content MUST conform to the date-time BNF rule in [RFC3339]. Note this requires an uppercase letter T between the date and time sections. Secondly, *all* RFC3339 date-times are compatible with the 4 specs mentioned, so the wording of the second paragraph ("As a result...") is a bit strange, since it's not as a result of anything we've done. Just say "Date values expressed in this way are also compatible with...". Graham [1]http://www.imc.org/atom-syntax/mail-archive/msg13116.html