Re: [netconf] IANA templates for algorithms

[Kent Watsen <[e-mail]>]

Hi Lada,

> Strictly speaking, the XSLT stylesheet is still intended only for producing the initial revision of the module.

Apparently  my expectations got ahead of me.


> Re-running the stylesheet on a later revision of the registry would essentially work, only the revision history would be lost. I don't see any easy workaround to this.

Perhaps pass the YIN of the “current” YANG module into the XSLT script too?


> It is up to IANA (we can help them) to decide whether the YANG module will be updated as before, e.g. by manually adding new enums, identities etc., or whether they use the stylesheet and then manually edit the revision history.

But it’s still a manual YANG-editor level effort, which is new (I assume) to IANA.


> Maybe it would be useful to write an informational RFC suggesting a workflow for these cases.

Don’t think that there is a patten just yet.  Your draft’s use XSLT to create the initial revision makes sense only in that the registry already exists and new registrations may occur while your draft is being published.  

In crypto-type's case, the intention is to create a brand new registry/YANG-module, so we could just as easily put the initial YANG module into the draft, negating the need for IANA to run `xsltproc` at all.

If IANA doesn’t run XSLT for subsequent updates, I think that we (NETCONF) need to question if a registry is even needed.   The motives behind defining a registry are 1) to accommodate IANA and 2) to make it seem more official/interesting to the Sec folks (since “YANG” appears to be uncool/taboo?).  But if IANA is okay editing the YANG, and the Sec folks are still uninterested, then we could skip creating the registry altogether.

That said, I feel that creating a, e.g., “base algorithms” registry is best, as it allows for potential future/other use cases, and I’m hopeful/optimistic that we can get the Sec folks onboard.


Kent // contributor