[rfc-i] xml2rfc issue

[brian.e.carpenter at gmail.com (Brian E Carpenter)]

On 04/02/2016 21:24, Erik Wilde wrote:
> hello carsten.
> 
> On 2016-02-04 00:10, Carsten Bormann wrote:
>> Interesting.  <eref target="http://www.tzi.org">TZI</eref> works with
>> xml2rfcv2 as specified in RFC 7749 in my little kramdown-rfc test set:
>> It translates into
>> ... TZI [1] ...
>> and a (weirdly named) section under References:
>> 7.3.  URIs
>>     [1] http://www.tzi.org
>> For a while, I have been reluctantly believing that this was actually
>> the intended behavior, and it is consistent with RFC 7749.
> 
> this is what i remember. for example, https://github.com/dret/I-D/blob/master/Published/xml-patch/draft-wilde-xml-patch-10.txt
> is an older draft (june 2014), and i don't know what version of xml2rfc i was using, but as you can see, it did work at that
> point in time in the same way as you are describing.
> 
>> So it seems there need to be specific circumstances to trigger the bug
>> below.  My test set uses symrefs for the biblio references, but it still
>> "works" if I switch that off, with the little flaw that both the biblio
>> references and the URI references are numbered starting from [1].  Hmm,
>> kramdown-rfc generates separate normative and informative references...
>>   (How did you switch that off?)
> 
> i am doing things very simply: i write the XML directly, and then i convert it via http://xml2rfc.ietf.org/.
> 
>> $ xml2rfcv2 --version
>> 2.5.0
> 
> http://xml2rfc.ietf.org/cgi-bin/xml2rfc-version reports 2.5.1.dev0
> 
>> (And, yes, I can reproduce the bug from draft-wilde-registries-01.xml. Hmm.)
> 
> not sure if i can help with better explaining the situation. this used to work as i thought it should, and now it doesn't
> anymore. i am not sure which version triggered the change in behavior, but as shown above, it did work roughly 1.5 years ago.

Yet I recall it failing about 7 years ago; so it's a very data-dependent bug.

Anyway: let's get the spec right in xml2rfcv3!

   Brian