Tim Ferguson wrote:
"Jamie Collins" wrote in
oups.com:

Just another thought: alarm bells ring in my head when I see a
single
column table.

Do you have any authoritative reference for these alarm bells? I
don't
know of any reason within R theory to avoid one-column tables, or any

other kind of all-PK tables.


I made it clear: just another thought and the alarm bells, both from
within *my* head g. But if you are looking for me to research
authorities for you vbg, I can recommend googling the exact phrase,
"When the attribute has a small set of values that do not change much
over time" and see who you get.

This design could work if you _knew_ that the colours would _never_,
_ever_ change.

You wouldn't want to do it too often, granted, but why 'never ever'?
Constraints can be dropped and redefined, data can be modified, life
goes on.

It is used sometimes, for example with Gender; but even
then it's something of a short cut.

Don't be afraid of the S word: google ISO sex codes (you may turn up
the same authority as before). What is the proper long way round of
doing it, then? Have a table of sex codes with a candidate key and
cascading FKs?

Jamie.

--