[AFS3-std] Re: abort packet format

[Tom Keiser]

On Wed, Mar 9, 2011 at 2:05 PM, Andrew Deason <adeason@sinenomine.net> wrote:
> On Wed, 9 Mar 2011 12:41:41 -0600
> David Boyes <dboyes@sinenomine.net> wrote:
>
>> >> What about a human-readable message?
>> >> And if we do that, what about internationalization?
>> >
>> >Maybe.
>>
>> Please, no. That's user-level stuff, not on the wire stuff. Translate
>> the message when you log it, not in the packet.
>
> Then you have to standardize a new code and upgrade the clients before
> they can make sense of it. If this stuff is only for human-readable
> information, there's no machine processing to be done on it, so it makes
> sense to just send the string.
>

Strongly disagree.  Even after disregarding aesthetic arguments,
internationalization pretty much dictates that directly sending
strings is a non-starter.  Error code lookup tables exist for a
reason: if you think it's sufficiently important that clients be able
to interpret error codes that were standardized post-build, then I
would recommend this happen via an out-of-band mechanism (e.g., a
distributed error lookup service)...

Regards,

-Tom