Re: lastVerified: optional vs. extension

[Joe Abley <jabley@isc.org>]

On Friday, Dec 6, 2002, at 10:25 Canada/Eastern, Klaus Malorny wrote:

> Error codes are not extensible, with the consequence that some 
> existing registries put additional error codes in the free text 
> elements (which are not thought to be parsed).

We have spent time scratching our heads trying to find error codes to 
describe particular conditions in our EPP server implementation, and 
the idea of an extensible structure for error codes corresponding to 
the currently-documented functionality seems like a useful one.

There is also the issue that a future object mapping might introduce 
services with error conditions that are just not accommodated by the 
currently-specified set of error codes. This might lead to subversion 
of free-text error fields (as Klaus described) into machine-readable 
subcodes, or to the reinvention of much of the base EPP functionality 
in other protocols rather than reusing the work that has already been 
done here.

Perhaps a structure where a generic, fixed and base set of error codes 
could be extended with sub-codes that are particular to a server 
implementation, or particular to an object mapping specification, might 
do the trick. Clients unfamiliar with particular server sub-codes could 
happily ignore them, whereas those clients which have a need to obtain 
more fine-grained information could parse them.

If this idea seems interesting, I would be happy to come up with some 
corresponding text.


Joe