[Tools-discuss] Bibxml7: anchor normalization, xml2rfc: caching with anchors

[Carsten Bormann <[e-mail]>]

I just tried to work around a problem in kramdown-rfc’s “stand_alone: false” mode (this is where kramdown-rfc doesn’t do the reference handling by itself but instead generates entity declarations like it were 1986).

I noticed that

https://xml2rfc.tools.ietf.org/public/rfc/bibxml7/reference.DOI.10.1145_1282427.1282421.xml?anchor=foo

generates

<reference anchor="FOO">

Hmm, ID/IDREF are case-sensitive in XML, so this doesn’t quite work for an <xref target=“foo”/>.
➔ bug1

Independent of what anchor I specify, xml2rfc creates a cache file called

reference.DOI.10.1145_1282427.1282421.xml

So if I have another document that does 

https://xml2rfc.tools.ietf.org/public/rfc/bibxml7/reference.DOI.10.1145_1282427.1282421.xml?anchor=bar

Xml2rfc still uses the cache entry with 

<reference anchor=“FOO">

➔ bug2

Bug 1 actually has another interesting facet:

https://xml2rfc.tools.ietf.org/public/rfc/bibxml7/reference.DOI.10.1145_1282427.1282421.xml?anchor=DOI.10.1145_637201.637236

creates

<reference anchor=‘DOI101145_637201637236' >

Whoa, what happened to my dots?

➔ bug1b

So can we make anchor normalization on the bibxml server less aggressive?
No upcasing (bug1), no removal of valid ID/IDREF characters (bug1b).

Also, we’d need to fix the caching in xml2rfc (bug2).
(Kramdown-rfc’s [1.2.12, unreleased] cache file names for URLs with anchors currently look like this:
reference.DOI.10.1145_1282427.1282421--anchor=foo.xml
)

Grüße, Carsten