Closed bernhardreiter closed 2 years ago
@bernhardreiter: Where did you get the information from that it should fulfill all ISO 8601 profiles?
From ROLIE RFC
https://datatracker.ietf.org/doc/html/rfc8322#section-6.1.3 does not say it explicitely.
https://datatracker.ietf.org/doc/html/rfc8322#section-7.4 mentions is explicitely
urn:ietf:params:rolie:property:content-updated-date
The "value" attribute of this property is a text representation
indicating the date that the content, referenced by the "src"
attribute of the Entry's "atom:content" element, was last updated.
This date may differ from the updated date of the ROLIE Entry
because updates made to the content and to the ROLIE Entry are
different events. The date MUST be formatted as specified in
[[RFC3339](https://datatracker.ietf.org/doc/html/rfc3339)].
But RFC3339 is a normative reference, we believe it must be [RFC3339] which the code enforces.
Well, https://datatracker.ietf.org/doc/html/rfc3339#section-5.6 provides z
and Z
… as that simplified profile of ISO-8601 modern RFCs should use :wink:
@bernhardreiter: Where did you get the information from that it should fulfill all ISO 8601 profiles?
Good catch, I've checked the CSAF-2.0 documentation, which does not explicitely refer to datetime standards, so I'd assumed that in absence of mentioning it, ISO 8601 would be a real-world expectation as it is a normative reference. I've made the mistake of not checking the ROLIE standards! Thanks for pointing it out.
@sthagen the problem is not the Z
, but the missing ":" time-second
, which is mandatory in
partial-time = time-hour ":" time-minute ":" time-second
@s-l-teichmann we could improve that checker message, I think, but this is worth to be handled elsewhere.
To be more precise the ROLIE specification lists RFC3339 as normative reference, which we interpret in the sense that all datetime values MUST use that format. (Even if it is only mentioned explicitely in section https://datatracker.ietf.org/doc/html/rfc8322#section-7.4 for The "rolie:property" Extension Point. (I've edited @s-l-teichmann comment above accordingly.))
As followup I've suggest that CSAF moves to RFC3339 with UTC only in https://github.com/oasis-tcs/csaf/issues/469
Here is a checker run, which encountered a ROLIE document with
which is a valid ISO 8601 format (see second paragraph in https://en.wikipedia.org/wiki/ISO_8601#Combined_date_and_time_representations for an example) and returned a
Loading ROLIE feed failed: parsing time "2022-02-22T00:00Z" as "2006-01-02T15:04:05Z07:00": cannot parse "Z" as ":".
Expectation: this format is accepted.