Upload
stacy-gay
View
21
Download
1
Embed Size (px)
DESCRIPTION
DCHK-05 meets Occam's Razor Marcos Sanz [email protected] 67th IETF, San Diego November 7, 2006. Advancing DCHK. Issues discovered while trying to upgrade our DCHK implementation from -04 to -05: vs Relevant "domain times" Lameness dateTime i18n - PowerPoint PPT Presentation
Citation preview
CRISP WG 2006/11/7
DCHK-05 meets Occam's Razor
Marcos [email protected]
67th IETF, San DiegoNovember 7, 2006
CRISP WG 2006/11/7
Advancing DCHK
• Issues discovered while trying to upgrade our DCHK implementation from -04 to -05:– <domainVariant>– <status> vs <enhancedStatus>– Relevant "domain times"– Lameness– dateTime i18n– Nits: wording, coherence, examples,
references, typos
CRISP WG 2006/11/7
Issue 1: <domainVariant>
• Assimilated from DREG• maxOccurs="unbounded" is scary in a
lightweight service• Potential high amounts of domain variants
don't play well with LWZ• Conceptually well placed in DREG(2), but not
in an availability service
CRISP WG 2006/11/7
Issue 2: <status> vs <enhancedStatus>
• Both can appear in a <domain> result object• However, <enhancedStatus> is meant to be a
superset of <status>• Further, <enhancedStatus> is more extensible
and flexible• So why keeping <status>? Backwards
compatibility?
CRISP WG 2006/11/7
Issue 3: Relevant "domain times"
• Current draft:– initialDelegationDateTime– lastDelegationModificationDateTime
• That is a mistake, it should be:– initialDelegationDateTime– createdDateTime
• Plan to add expirationDateTime• Plan to add lastDatabaseUpdateDateTime
CRISP WG 2006/11/7
Issue 4: Lameness
• <enhancedStatus> includes <lame>, which is not an instance of <enhancedStatusType>
• It has been assimilated from DREG2• It is ill-defined in this context: it is an assertion
about the lameness of a name server, not of a domain
CRISP WG 2006/11/7
Issue 5: dateTime Internationalization
• Unnecessary restriction in section "Internationalization Considerations":"The [...] element contains the XML Schema data type
'dateTime'. The contents [...] MUST be specified using the 'Z' indicator for [...] UTC"
• No guidelines about that in BCP70 or RFC3076• Probably only meant to recommend "Z" vs "z"
(v RFC3339)
CRISP WG 2006/11/7