Re: finding gaps in temporal data

I have a fact table that looks like this:

dim1_fk bigint,

time_fk bigint,

tstamp timestamp without timezone

value numeric(16,2)

The tstamp column is completely redundant to a colume in the time_utc table, but I like to leave it there for convenience when writing ad-hoc queries in psql - it allows me to skip the join to a time dimension table.  The fact table is actually partitioned into 1-month child tables, with check constraints on both time_fk and the tstamp column, since there is a 1:1 relationship between those columns.

dim1_fk refers to a dim1 table which has two varchar columns we care about:

dim1_pk bigint,

label1 varchar,

label2 varchar

the time_utc table contains the usual time dimension columns, and I've structured the primary key to be an integer in the form YYYYMMDDHH24mm - so 2011-06-15 15:35 has a primary key of 201106151535 and there is a row every 5 minutes. All data in the fact table is assigned to a given 5 minute window. There is a row in the time_utc table for every possible time value, regardless of whether there is data in a fact table for that 5 minute interval.  For our purposes, we only need care about 2 columns

time_pk bigint,

tstamp timstamp without time zone

I'm looking to run a report which will show me any gaps in the data for any label1/label2 pair that appears in the dim1 table - there are 0 or more rows for each label1/label2 combination in each 5 minute window and I don't actually care about the duplicates (all my queries aggregate multiple rows for a given timestamp via the avg() function).