Re: RFC-2445: ABNF/prose mismatch on parameters

[<swansma@xxxxxxxxxxxxx>]

> 
> In any case, the issue of ABNF vs prose discrepancies needs to be 
> resolved.  I think that we agreed that any discrepancies should be 
> resolved in favor of the text.  However, as I noted originally this may 
> make the ABNF really really nasty and I would be against any _signficiant_ 
> nastyness or bloating.

Me too.

> WRT:
> >Is it so bad that we parse an object, then
> >use code to test its validity?
> 
> BTW: There is no way w/any amount of coding (no matter how good you are) 
> that discrepancies between ABNF and text can be rectified pre or post 
> parsing.  Validity checking, etc can be done separately from "Is Due and 
> Duration allowed on the same vEvent and if so, how many times may each 
> occur?" etc.   If your starting point (the ABNF) is inaccurate when 
> compared w/the descriptive text then your starting point is flawed and it 
> affects everything else you do later on.
> 
Yes, when there are discrepancies. I had failed to find the discrepancies
in the example you had given and assumed that the request was simply to change the
 prose into ABNF.

To clarify: I didn't state code could rectify discrepancies, that would
just be too illogical:-) I was simply voting to keep the prose (fixing any
discrepancies of course) because it has been easy for me to meet the
requirements stated in the prose with some code.

I did miss the post that explained exactly which prose was at odds with
what ABNF. Could someone please post it or email it to me? I'd appreciate
it.

Thanks.