Enforcing ISO-3166-1

[Shane Kerr]

On 2001-02-21 14:22:13 +0000, ginny listman wrote:
> 
> > Well, perhaps you could post a sample list here, and we can see if
> > the entries are indeed obvious to the readers.  From my experiences
> > with this topic, I feel that you can safely "auto-correct" most of
> > the problems with almost no mismatches.  The examples you gave of
> > Brasil and UK, for instance, are extremely common.  Things like
> > Britian are also fairly common - not much ambiguity there, just
> > wrongness.  ;)
> 
> My concern is not of ambiguity, rather a data management nightmare.
> For example, we have UK, Great Britain, England, etc.  None of which
> are listed in ISO 3166.  What is there is United Kingdom, GB and GBR.
> I can think of at least 5 ways of expressing US, that are not
> "standard".  Are we suppose to list to dbwg (or some other list) to
> make sure we've capture all the common names?  Or just see what common
> errors we accumulate?  Why can't we just educate the members and
> inform them about ISO 3166?  If we accept a template with U.S.A., are
> we promoting the ignorance of an ISO standard?

Given a choice between smarter, but imperfect, software and the
impossible and unending task of attempting to educate the world why not
choose making life easier on users and yourself? 

This particular problem has the wonderful property of dealing with
largely static data (country codes), plus having having a wonderful
source of sample errors easily avaible (the existing ARIN database).
Most of the work only has to be done once (and in fact I did this very
thing once before), requiring only minor (if any) subsequent
modifications.

<philosophy>
It's the job of the developer to attempt to anticipate common errors and
correct them in an intuitive way.  Programs should be made to work for
people, not people for their software.
</philosophy>

> > The real answer to this is of course a web (or other GUI) interface,
> > so the user can pick from a list.  ;)
> 
> Agreed.  Or intention is to have a web based template.  However, not
> every member would be willing to submit via the web.  Therefore, we
> will need to address text-based templates.

In a world with both a web interface and an e-mail interface, then I am
totally in favor of strict rules.  The target audience for such an
interface is expert users and automated tools, both of which would
probably prefer the parser to do what you tell it to, not what it thinks
you mean for it to do.

My 0.022 Euros...

Shane