[rfc-i] IETF RFC format <-> W3C pubrules

Phillip Hallam-Baker hallam at gmail.com
Fri May 11 11:11:34 PDT 2012

OK, this is getting a little far off the course but what you are
highlighing is that compiling citations is really the sort of project
where a network effort can really pay off.

And don't just think of this for citations either. The big stumbling
block for semantic Web as I see it is a chicken and egg problem.
People don't do metadata when it isn't used. People can't use metadata
when it isn't present. When the CD spec was first developed, they had
a scheme there to encode the track names. It wasn't ever used until
the rippers came along and so almost nobody bothered to fill it.

A group effort does not need to be a permanent fixture, it could
merely be transitional, until people start filling in the metadata

Let us imagine that there are multiple services that host citation
databases and I can open an account at any one I choose (that accepts
me). The services may or may not share data between each other.

A citation generator tool would go over the source of a document and
attempt to map any form of citation it understood (URL, paper name,
etc) to a citation by querying the service(s) that the user had

Depending on the configuration of the tool I could then complete the
references by entering any remaining unresolved references in my
document or by logging into my account and updating them there. In
either case I could choose to make my citation data available to other
people or not. Having the 'or not' piece being essential since keeping
grotty data out of a database is always more important than simply
aggregating poor data.

A tool could be made to work without accounts of course, but accounts
make things much easier as that allows the citation search to be
personalized. Expanding RFC822 to a citation to the IETF doc is pretty
obvious for someone like me but may not be for someone in a different

Using accounts also makes it possible to write a tool that runs
without requiring user input. I really loathe tools that try to lock
me into their particular workflow. Like all those tools that offer to
download pictures from a flash card and save them on the hard drive
that then try to force you to look at the pictures and catalog them at
the same time.

Rather than rush to a spec though, I think this should be thought of
as something we might build into a spec production toolset for
internal SDO use but with a view to opening it up for wider use later

On Fri, May 11, 2012 at 12:38 PM, Sandro Hawke <sandro at w3.org> wrote:
> On Wed, 2012-05-09 at 08:57 -0400, Phillip Hallam-Baker wrote:
> ...
>> For example, it should be possible to cut an paste a citation from one
>> document to another in such a way that tools are able to reformat it
>> to apply whatever deranged nonsense of a citation format is required
>> at the other end. I don't see that as existing.
>> Pretty much every tool there is to manage citations sucks. I have
>> tried end note and it sucks because it is an afterthought. The
>> citation handling in Word is stovepiped to a few formats that are all
>> stupid and few other things bother at all.
>> It really should not be difficult, A 'database' of citations should
>> require no more than an HTML document with a list of citations.
> Just following on this thread, here's a somewhat extreme strawman,
> which probably shows my W3C bias.
> How about having the citations and references section all generated
> automatically?  Have the generator use:
>  1. The HTML links going out of the document
>  2. Some indicator of whether a particular outbound link target
>     should not be considered a citation, or if it is one, whether it
>     should be considered non-normative.  In HTML this could be done
>     with a class attribute; alternatively, it could be done via a
>     list of URLs somewhere in the document metadata.
>  3. Some URL->Citation databases.  Ideally, all the citation
>     information could be found by dereferencing the URL and looking
>     at standard metadata.  Until then, some databases can backfill
>     that information.  I'd hope we could suffice with one of these
>     databases at W3C and one for the RFCs.  I'd hope they could use
>     the same format.
> In the case of wanting to cite something which has no plausible URL, I
> suggest one be created, giving some useful information about the cited
> item for people who do not have access to a suitable library.
>   -- Sandro

Website: http://hallambaker.com/

More information about the rfc-interest mailing list