Question

I have an sqlite database structured as follows:

CREATE TABLE IF NOT EXISTS Patient 
( PatientId INTEGER PRIMARY KEY AUTOINCREMENT );

CREATE TABLE IF NOT EXISTS Event 
( 
PatientId INTEGER REFERENCES Patient( PatientId ),
DateTime TEXT,
EventTypeCode TEXT,
PRIMARY KEY( PatientId, DateTime, EventTypeCode )
);

CREATE TABLE IF NOT EXISTS Reading 
( 
PatientId INTEGER REFERENCES Patient( PatientId ),
DateTime TEXT REFERENCES Event (DateTime),
EventTypeCode TEXT REFERENCES Event (EventTypeCode),
Value REAL,
PRIMARY KEY( PatientId, DateTime, EventTypeCode )
);

I insert a Patient with Id #1

then I run:

INSERT INTO Event (PatientId, DateTime, EventTypeCode) VALUES (1, '2011-01-23 19:26:59', 'R')

which works

then I run:

INSERT INTO Reading (PatientId, DateTime, EventTypeCode, Value) VALUES (1, '2011-01-23 19:26:59', 'R', 7.9)

and it gives me a foreign key mismatch. Patient Id is ‘1’ in all cases, and the datetime and typecodes match in the 2nd and 3rd queries. I do not understand what is mismatching, but I’m a bit new to actually defining foreign keys and i do not know what I am doing wrong.

Answer 1

I’m not familiar with SQLite but a little Google’ing turned up this. The documentation says

If the database schema contains
foreign key errors that require
looking at more than one table
definition to identify, then those
errors are not detected when the
tables are created. Instead, such
errors prevent the application from
preparing SQL statements that modify
the content of the child or parent
tables in ways that use the foreign
keys. Errors reported when content is
changed are “DML errors” and errors
reported when the schema is changed
are “DDL errors”. So, in other words,
misconfigured foreign key constraints
that require looking at both the child
and parent are DML errors. The English
language error message for foreign key
DML errors is usually “foreign key
mismatch” but can also be “no such
table” if the parent table does not
exist. Foreign key DML errors are may
be
reported if:

• The parent table does not exist, or
• The parent key columns named in the foreign key constraint do not exist,
  or
• The parent key columns named in the foreign key constraint are not the
  primary key of the parent table and
  are not subject to a unique constraint
  using collating sequence specified in
  the CREATE TABLE, or
• The child table references the primary key of the parent without
  specifying the primary key columns and
  the number of primary key columns in
  the parent do not match the number of
  child key columns.

I suspect you might be running into #3 in that list.

Also, while other DBs might support using a non-unique index as a foreign key reference, (see answers here), it’s a bad design choice in my opinion. I would restructure so that either

1. Reading.PatientId references Event.PatientId so that the complete composite key from Event is referenced by Reading or,
2. Add an EventId auto-increment, primary key to the Event table and use that as the foreign key in the Reading table (so that you only have EventId and Value under Reading and you can get the PatientId, DateTime, EventTypeCode out of Event).

I’d suggest #2 so that you can avoid the redundancy of PatientId, DateTime and EventTypeCode in both Event and Reading.