Checking Query Results Using RowFixture

This style of test table is best for checking the results of queries. For the sake of discussion, let's call each of the items that you get back from a query a record. A RowFixture lets you test that you get back exactly the set of records from a query that you expect to get.

Here is a simplistic example. Let's say that we have asked our application for the first five prime numbers. We want to make sure we got them all.

fitnesse.fixtures.PrimeNumberRowFixture

prime

3

2

5

7

11

This test table style does not read the way that a ColumnFixture style test table does. In this case, each cell in the prime column represents a key that identifies one of the records we expect to get back (in this case, a prime number). And the entire set of rows of data represent the output we expect to get back: no more and no less than that exact set of records (though they need not be in that exact order).

Hit the test button to run this test. OK, we got them all. For an explanation of the fixture code for this RowFixture example, see RowFixtureCode.

Missing Records

The following table shows what happens when you ask for a record that doesn't exist. (Note the extra 12.)

fitnesse.fixtures.PrimeNumberRowFixture
prime
2
3
5
7
11
12

Surplus Records

And this table shows what happens when you don't ask for a record that is in the query results. (Note the 11 is missing.)

fitnesse.fixtures.PrimeNumberRowFixture

prime

2

3

5

7

Testing Fields in Returned Records

Often you want to see that you got a specific set of records back, and you want to test certain fields in the returned records. For this you use the same question-mark syntax we saw used for output columns in ColumnFixture style tables.

For example, lets say that we have a database of employee records. We want to test that the employees were paid correctly. The pay? column in the table below specifies the exact amount we expect to get back for the pay field of each returned employee record:

fitnesse.fixtures.EmployeePayRecordsRowFixture
id	pay?
1	1000
2	2000

Incorrect Output

Here's what it would look like if the pay was not correct for the employee whose id is 2. It looks just as it does for incorrect output values in a ColumnFixture table cell:

fitnesse.fixtures.EmployeePayRecordsRowFixture
id	pay?
1	1000
2	3000

Missing or Extra Records

And regardless whether you check aspects of queried records, FitNesse shows you whether any are missing or extra:

fitnesse.fixtures.EmployeePayRecordsRowFixture
id	pay()
1	1000
5	5000

To see the fixture code for these tables, see RowFixtureCode.

Querying Using Multiple "Keys"

Sometimes we need more than one piece of data to uniquely identify a record.

Suppose we have records for the last three months' worth of paychecks. Each paycheck is uniquely identified by the employeeId and the date. We want to make sure that the checks were generated properly. We use a RowFixture table to ask for a set of checks identified by both employeeId and date:

fitnesse.fixtures.PayCheckRecordFixture
date	employeeId	name	pay()
3/1/03	1		1000
3/1/03	2		2002
4/1/03	1		1015
4/1/03	2		2003

Notice, once again, that the order of the table rows doesn't matter. For example, this table also works fine for that same query:

fitnesse.fixtures.PayCheckRecordFixture
date	employeeId	name	pay()
4/1/03	1		1015
3/1/03	1		1000
4/1/03	2		2003
3/1/03	2		2002

Blank Fields

Notice the name field. We provided the field name, knowing that it was part of each record, but we provided no expected values in that column. As a result, FitNesse did not make this field part of the key for finding records, but merely filled in the values found for that field in each record, so we could see them. Here's what would have happened if we filled in the name column correctly for some cells, and incorrectly for another.

fitnesse.fixtures.PayCheckRecordFixture
date	employeeId	name	pay()
3/1/03	1	Bob	1000
4/1/03	1	Bob	1015
3/1/03	2	Bill	2002
4/1/03	2	Biff	2003

And here's what would happen if a row were missing or surplus.

fitnesse.fixtures.PayCheckRecordFixture
date	employeeId	name	pay()
3/1/03	1		1000
4/1/03	1		1015
3/1/03	2		2002
4/1/03	3		2003

Summary

• RowFixture tables are designed for checking that you get back exactly the set of records you want from a query.
• FitNesse will cause the test to fail if the query had any missing or surplus records.
• You can also test fields in the returned records.
• If you provide a field name column, but provide no expected values for it, then FitNesse will fill in the returned values so you can see them.

Learning More

To learn about the fixture code for the above examples, see RowFixtureCode.
Also check out the other TestTableStyles.