RE: Alpha Five compared to Access
Cal
....I think the A5 version of Access's "Unique" index is to set the Validation field rule for the field to "Value of field must be unique".
Steven
Yes, this is the "workaround" method, and when you hang out on this forum, you find out that it means the following..
a) You must define the field rules in every set that
allows update to the table, as well as the table itself..
b) You should define the same rule for each field (2,3,4)
that makes up the composite key, because of the
possibility of data entry not going in "order".
c) Child table field rules have in some cases caused
anomalies in entering parent table data.. triggering
where they shouldn't (is that all fixed now ?, I am not
sure).. there is even another possible issue of
triggering involving fields that can be blank/zeros
Clearly nowhere near as simple, clear and elegant than OS or language-level unique keys... not really close..
Cal
If you want a uniqueness test that is more complex, you can even define "Value of expression must be unique" which allows you to combine the values in two or more fields for the uniqueness test.
Steven
Agreed.. this is a nice powerful feature..(and it may be much better than the competitor's equivalent, I dunno)
Would be nice to also see it implelmented at the language/table level, along with true unique keys.
Cal
The only purpose I can think of right now for a unique index in A5 is to use it for a lookup.
Steven
Primary keys are the basic "key" to data integrity, and on active data they are almost invariably unique indexes, Anything that can compromise the integrity of unique indexes must be watched for very carefully. Of course if you embrace the workaround method, then the proper, elegant method lacks purpose :-)
Jermay
OR, you could create a unique index so each city would only be listed once - much easier.
Steven
That is not the definition of a unique index, it is a (programming or languge) "summary field" type of function and should be defined as such.. a very limited function best handled in report or browse level.. calling it a unique index is, as Jeremey essentially pointed out, kicking against the goads of accepted database terminology, and clearly able to cause developers a lot of grief unless they sleep by this forum :-) when the "duplicate key" data slips through their "unique index" unknowlingly and invisibly...
I know what I am saying here sounds a little harsh, and at first I was gonna largely accept your view, but the more I thought of it, the more I realized that missing true unique indexes (and labeling something unique that is not) is not something that should be sluffed off, either in comparisions, or in requests to Alpha for future enhancement...
Shalom,
Steven
Cal
....I think the A5 version of Access's "Unique" index is to set the Validation field rule for the field to "Value of field must be unique".
Steven
Yes, this is the "workaround" method, and when you hang out on this forum, you find out that it means the following..
a) You must define the field rules in every set that
allows update to the table, as well as the table itself..
b) You should define the same rule for each field (2,3,4)
that makes up the composite key, because of the
possibility of data entry not going in "order".
c) Child table field rules have in some cases caused
anomalies in entering parent table data.. triggering
where they shouldn't (is that all fixed now ?, I am not
sure).. there is even another possible issue of
triggering involving fields that can be blank/zeros
Clearly nowhere near as simple, clear and elegant than OS or language-level unique keys... not really close..
Cal
If you want a uniqueness test that is more complex, you can even define "Value of expression must be unique" which allows you to combine the values in two or more fields for the uniqueness test.
Steven
Agreed.. this is a nice powerful feature..(and it may be much better than the competitor's equivalent, I dunno)
Would be nice to also see it implelmented at the language/table level, along with true unique keys.
Cal
The only purpose I can think of right now for a unique index in A5 is to use it for a lookup.
Steven
Primary keys are the basic "key" to data integrity, and on active data they are almost invariably unique indexes, Anything that can compromise the integrity of unique indexes must be watched for very carefully. Of course if you embrace the workaround method, then the proper, elegant method lacks purpose :-)
Jermay
OR, you could create a unique index so each city would only be listed once - much easier.
Steven
That is not the definition of a unique index, it is a (programming or languge) "summary field" type of function and should be defined as such.. a very limited function best handled in report or browse level.. calling it a unique index is, as Jeremey essentially pointed out, kicking against the goads of accepted database terminology, and clearly able to cause developers a lot of grief unless they sleep by this forum :-) when the "duplicate key" data slips through their "unique index" unknowlingly and invisibly...
I know what I am saying here sounds a little harsh, and at first I was gonna largely accept your view, but the more I thought of it, the more I realized that missing true unique indexes (and labeling something unique that is not) is not something that should be sluffed off, either in comparisions, or in requests to Alpha for future enhancement...
Shalom,
Steven
Comment