Re: USMARC and postgresql?

Andrew Sullivan <sullivana@bpl.on.ca> writes:
> This is way hard.  The reason is simple: MARC allows all of the following:

> 1.    Multiple instances of the same field.
> 2.    Variable length of field.
> 3.    Unpredictable numbers of completed fields.

I would think you could handle this by splitting an entry into multiple
records linked by a common ID value.  In barest bones:

    itemid int,
    fieldtype int,
    fieldvalue text

and you do a SELECT ... WHERE itemid = X and fieldtype = Y to retrieve
all the instances of a particular field.

Still, looking around for someone who's already done it seems like a
fine idea ;-)

            regards, tom lane