Re: Normalization Question

>> ...how would phone# subparts (ie Extentions, Hours of operation,
>> Alternate number, etc) be stored without redundancy?
>
> add another field called INFO and one called INFO.TYPE
> and have a flag in the type field like E\H\A ...
>
> H]C]C
> 1231231234]1231234]1231231234
> E\H]E\A]E
> 123\3800-18000]123\1231231234]123

The above is interesting as I haven't worked with MVDBs.

// Below script models the above with XDb2 (CREATE *.name = +John

_       & it.cls = person
_       & it.phone# = (CREATE *.name = +1231231234
_                             & it.cls = phone#
_                             & it.cls = home#
_                             & it.ext# = +123
_                             * it.hours = 0800-1800)
_       & it.phone# = (CREATE *.name = +1231234
_                             & it.cls = phone#
_                             & it.cls = cell#
_                             & it.ext# = +123
_                             & it.alt# = +1231231234)
_       & it.phone# = (CREATE *.name = +1231231234
_                             & it.cls = phone#
_                             & it.cls = cell#
_                             & it.ext# = +123))
(CREATE *.name = +123
_       & it.cls = robot)}

// Find a person whose cell# has the ext# 123. // Finds John.
(SELECT *.cls=person & *.phone#=(*.cls=cell# & *.ext#=123))

// Find phone# that have ext# 123.
// Finds all three of the above phone#s. (SELECT *.cls=phone# & *.ext#=123)

// Finds things named 123.
// Find the ext# 123 and robot 123.
(SELECT *.name=123)

> As far as distribution, fulfillment and the like,
> it's really the best system going.

Yes, I would agree that each modelling method/system has it's unique set of adv/disavantages making it especially well suited for a certain range of applications. Received on Thu Feb 03 2005 - 17:58:32 CET