Posted by Bent Stigsen on 05/18/06 20:07

Jerry Stuckle wrote:
> Kenneth Downs wrote:

[snip]


I have a few comments to add to Jerry's...


>> How much code will you write to support the new value? Any? Do you know?
>>
>
> That's his (and my) point. None.

In this case, I will say so too. All code up to beneath the userinterface,
can be made indifferent to any extra values without any real effort,
anything else would be bad coding.

Any userinterfaces, which must display this extra value, can be a different
story as it should look nice and be userfriendly. Meaning, if the
requirement changed from 8 values to a hundred, then a change of layout and
features would probably allways be needed. But that would mean poor prior
research.


>> How will you propagate your changes from test to production? Will your
>> collaborators know you did this? Do you have only one customer running
>> this code or several?
>>
>
> insert record in table with characteristics in production system.

Yes, indeed. Or just mail a sql-script to someone with access.


>> How is the new value documented, do you need to document it?
>>
>
> Self-documenting.

In this case, why document something that may be common knowledge within an
industry, when all they need to know is where to find the characteristics,
thingies or doodles of xxx. If they require "yellow windows" to be present,
they should know what it means.

In other cases it might need documentation, but has no importance to the
choice of table design.


>> Are there constraints on allowed values? Does the presence of the new
>> value affect others columns in other tables?
>>
>
> Whatever constraint the customer wants to put on it. And no, it does not
> affect other columns in the table.
>
>> Do any flags exclude other flags? Do any flags require any other flags?
>>
>
> Nope.

At least unknown.

Kenneth, if there were such constraints, it would be additional information
you would have to deal with too.


>> As I've said now three times in the thread, we don't know, we're just
>> playing hypotheticals. If and only if none of these questions matters,
>> than go with your cross-reference, because you are talking about user
>> data then. (In fact by definition you should not then be doing it, they
>> should
>> be putting it in). But if even one of them matters than you aren't done
>> yet.
>>
>
> You're adding a new CHARACTERISTIC. Whether or not they have wi-fi
> facilities
> has absolutely nothing to do with the color of their windows. That was in
> the original definition of the problem - these are independent
> characteristics.

Kenneth, it is not more hypothetical than a real situation for Niclas, who
started the thread. I don't consider the hypothetical speculations along
the way of much importance in this discussion. Niclas didn't mention
constraints between the characteristics, but we can play that he did, each
make an adjustment, and we could have the same discussion. If not mentioned
or not thought of, then indeed I wouldn't be done, but that is not a flaw
in the design, rather a change in the scenario which would be just as
relevant for you, requiring the same amount of work. When that work is
done, Jerry and I still got a system indifferent to the number of
characteristics.

*if* it was decided, that it should be user data (not sure what definition
you refer to), then I wouldn't have to lift a finger, because then there
would be a method for the user to modify the characteristics themselves.
But not something I would do unless required.


/Bent

[Back to original message]