[rfc-i] Better markup may get processed faster

Joe Hildebrand jhildebr at cisco.com
Tue May 29 11:58:51 PDT 2012


On 5/29/12 11:25 AM, "Joe Touch" <touch at isi.edu> wrote:

>> ABNF requires a skillet that is not as common in that population.

"skillset" obviously, sorry.

> Sure, but it too is often checked by people, since the tools tell you
> whether it's syntactically correct but cannot confirm it is semantically
> correct.

There are some more interesting checks that are easy, such as ensuring that
there are no rules that have been left undefined.  It's also possible to
generate a scanner from the ABNF, and test that against sample inputs.

>> Think about the cost of the people involved.
> 
> Right. Checking the entire document for spelling and grammar vs. one
> section that isn't in most RFCs.

Even more reason why the RFC editor might not hire for the skills required
to check ABNF.

>>> Optimizing for these sorts of things
>>> is a good way to end up with a
>> system that is complex enough to result in
>>> delays elsewhere too.
>> 
>> I wrote an ABNF parser in JavaScript over the weekend
>> (https://github.com/hildjj/node-abnf).  It wasn't that complex.
> 
> Sure - which you can cut/paste into; you don't need markup to find the
> ABNF in a doc.

I'd prefer my tooling to run these checks automatically every time I submit,
so I don't have to hunt through the doc to find all of the pieces of ABNF.

>> Could you give an example of the sorts of the delays you're worried about?
>> I'm concerned that this might have been an off-the-cuff statement without
>> any technical basis.
> 
> Ask the RFC Editor staff how much time they spend on grammer/spelling
> issues vs. ABNF and code verification. ISI ran that position for many
> years, in my division. I don't have detailed measurements, but I do know
> they spent a lot more time on grammar/spelling than most other things.

It's possible that both of those teams missed issues that would have been
trivial to fix by the author if they had been caught in AUTH48, then.  It's
also possible that those teams didn't spend much time on it because they
didn't have the skills to do so.

-- 
Joe Hildebrand



More information about the rfc-interest mailing list