[rfc-i] DTD breakage for reference IDs for 3GPP (bibxml5)

Julian Reschke julian.reschke at gmx.de
Thu Nov 7 16:20:56 PST 2013

On 2013-11-07 16:19, Julian Reschke wrote:
> On 2013-11-07 14:35, Tony Hansen wrote:
>> One of the items I've been looking at this week is what to do with the
>> bibxml5 references to 3GPP.
>> In a nutshell, the problem is:
>>      *) the 2629 DTD says that the anchor attribute for <reference> is
>> an ID
>>      *) an ID is required by the XML specs to start with an letter
>>      *) xml2rfc v1 ignored this and allowed anchors to start with digits
>> as well
>>      *) bibxml5 is *filled* with anchors that *all* start with "3GPP"
>>      *) xml2rfc v2 follows the DTD to the letter and generates errors for
>> all of these anchors
>> My first response to this was to simply change the 3GPP references to
>> something else. But what should it be changed to? "SDO-3GPP" works, but
>> none of our other SDO references (e.g., IEEE, ANSI, etc) use "SDO" as
>> part of their reference name. How about prepending "TR-" or "TS-" as
>> appropriate for 3GPP technical reports and technical specs? That's again
>> different from all other SDO's uses, where it's always the SDO name that
>> begins the reference.
>> Then, Sandy asked if we couldn't just make it allowable for 3GPP to be
>> used again? And that's where this message is coming from.
>> My questions are:
>>      *) What would the consequences be if we were to change the 2629 DTD
>> such that an anchor allowed digits to start the anchor?
> That validation doesn't check anchor uniqueness anymore.
> -1 for that.
> This seems to be a case that might be worked around by "vanity reference
> names".
> Best regards, Julian

Furthermore it means that code that generates HTML from XML will work 
around this, because it'll generate broken anchors otherwise.

Best regards, Julian

More information about the rfc-interest mailing list