Re: [regext] draft-ietf-regext-epp-fees-02.txt: currency error handling, command wildcard

[Bernhard Reutner-Fischer <[e-mail]>]

On 24 March 2017 11:05:44 CET, Thomas Corte <[e-mail]> wrote:
>Hello Bernhard,
>
>On 23/03/2017 16:04, Bernhard Reutner-Fischer wrote:
>
>> In my view 2004 is more on the protocol-layer, i.e. value supplied
>not in
>> curl -s -o-
>https://www.currency-iso.org/dam/downloads/lists/list_one.xml | \
>>  sed -n -e "/<Ccy>/s/.*>\([^<]*\)<.*/\1/p" | sort | uniq | tr "\n" "
>"
>> 
>> 2306 is a policy violation, i.e. may be a valid currencyType but is
>> not supported by your registry.
>
>I see, however it seems futile to have these two levels of checking...
>the server-supported currencies are obviously a subset of the globally
>available currencies, so any currency not present in the long list is
>bound to be rejected later on when the server checks for support.
>
>If at all, I'd probably rather like to see the restriction to existing
>currencies done via the schema, i.e. by an explicit enumeration of
>valid
>values in the XSD.

That's what normative references are for IMHO. You wouldn't want to update the fee RFC if someone invents a new currency.

You might want to throw an error if someone configures the server to use a new/typoed currency and always 2306 when receiving a currency outside of the configured enumeration, thus ignoring 2004 completely at runtime.

cheers,