Re: [xml2rfc] RFC Bibtex format doi numbering incorrect

[Julian Reschke <[e-mail]>]

On 2017-05-26 18:52, John Levine wrote:
> In article <[e-mail]> you write:
>> On 2017-05-25 21:49, Robert Sparks wrote:
>>> Forwarding to what I hope is the best list...
>>
>> Out of curiosity: why is it incorrect? Do we have a precise description
>> about what the format should be? Hopefully including RFC#s > 9999?
> 
> DOIs are opaque identifiers.  The DOI of any RFC is whatever is in the
> DOI field of the RFC Editor's database.  Any code that attempts to
> guess the DOI from the RFC number is broken.

Wow. Really? Why not just define the mapping precisely?

> The DOI gemeration code uses the editor database internal identifier
> as the last component of the DOI, because the code was easy to write.
> Those identifiers currently all look like RFCnnnn but they may change
> at any time without notice, since they're purely for the editor's
> internal use.  In retrospect, using them was a mistake.  I should have
> used something obviously opaque like a hash of the title and issue
> date for the DOI.

Right. So if they are for the editor's internal use only, what's the 
correct way to compute the DOI for a RFC then?

> In response to the question of what DOIs of RFCs past RFC9999 will be,
> it doesn't matter.  If your code tries to guess, it is still broken.
> The DOI is whatever the DOI is.  Don't guess, look it up.

Where? I thought you just said the database is for internal use only?

Once again: just define the mapping precisely.

Best regards, Julian