>-----邮件原件----- >发件人: netmod [mailto:netmod-boun...@ietf.org] 代表 Jürgen Sch?nw?lder >发送时间: 2022年3月24日 18:56 >收件人: mohamed.boucad...@orange.com >抄送: netmod@ietf.org >主题: Re: [netmod] TR: New Version Notification for >draft-boucadair-netmod-iana-registries-00.txt
>On Thu, Mar 24, 2022 at 10:44:42AM +0000, mohamed.boucad...@orange.com wrote: >> > It seems that later on you are actually changing what RFC 8407 says >> > although I am not sure I understand the reasoning. RFC 8407 >> > recommends to use enums if there is a single naming authority >> > allocating values and thus ensuring uniqness of values. I am not >> > sure in which sense the DOTS decision to use enums is not inline >> > with what RFC 8407 says. DOTS may have decided to go with enums for >> > space reasons and then this decision implies that values have to be >> > centrally allocated. Note that there are IANA registries that allow >> > distributed allocation of values and for thoses cases the RFC 8407 >> > recommendation to use identities still applies I think. >> >> [Med] I was referring to this part: >> >> If extensibility of enumerated values is required, then the >> "identityref" data type SHOULD be used instead of an enumeration or >> other built-in type. >> >But why do you think this statement of RFC 8407 needs any changes or different >interpretations? [Qin Wu] If my understanding is correct, Authors of document containing YANG data model face dilemma choices now, i.e.,whether 1.change enum into identities which avoid updating the module defining the enum in the future 2. Or stick to use enum and separate all enum types into IANA-Maintained YANG Modules, which also avoid updating the module, in addition, another benefit is to make sure the same registry maintenance for the module that get updated. For the second choice, here is one related discussion raised by Lada https://mailarchive.ietf.org/arch/msg/netmod/AILS-FptdNxMWFgSHiSjbwGc6tM/ both seems to work. >An enum implies that allocations can only be done by updating the module >defining the enum. If a WG goes for an enum because of efficiency concerns, >then the WG accepts that new allocations require an update of the module >defining the enum. If you >want multiple parties to allocate values, then the price is the overhead of >using identities. >The good old middle ground are numbers spaces where some of the space is >allocated for distributed allocations (and sometimes these allocations than >include things like enterprise IDs to reduce the likelyhood for collisions) >but then the YANG equivalent >for this is not just a plain enum. /js -- Jürgen Schönwälder Jacobs University Bremen gGmbH Phone: +49 421 200 3587 Campus Ring 1 | 28759 Bremen | Germany Fax: +49 421 200 3103 <https://www.jacobs-university.de/> _______________________________________________ netmod mailing list netmod@ietf.org https://www.ietf.org/mailman/listinfo/netmod _______________________________________________ netmod mailing list netmod@ietf.org https://www.ietf.org/mailman/listinfo/netmod
