Test case : Query Tool

A Simple Query Tool
===================

Informal specification
----------------------

The tool must support some implementation of tables.

A table has rows and columns.  The rows are numbered and the columns
are named.  Row-numbering may begin at either 0 or 1.  Column-names
must be unique within a table.

A table cell may contain arbitrary data (optionally including tables.)

The tool contains the following functions:

1)	table <- rows(table, selection-criterion)

A selection-criterion is a boolean expression over the columns of
the table.  The 'rows' function returns a table whose columns are
those of the input table and whose rows are just those of the input
table which satisfy the selection criterion.  For example,

	rows(employee, (age<40) and (salary > 20,000)

returns a subset of the table containing information for those
employees younger than 40 and earning more than 20,000.

2)	table <- cols(table, column-list)

The 'cols' function returns a table with the same number of rows
as the input table, whose columns are those in the column-list.
For example, the result of

	cols(employee, (age, salary))

contains just the two columns age and salary.

Together, these two functions can be used to simulate SQL 'select':

	select age, salary from employee where age < 40 and salary > 20,000

3) 	table <- order(table,column-list,direction-list)

The 'order' table sorts the input table according to the column-
list and the direction-list.  For example,

	order(employee, (age, salary), (asc, desc))

sorts the employee table by salary descending within age ascending.
This is similar to SQL's 'order by' clause:

	select * from employee order by age, salary(desc)

4)	table <- group(table, key-list, aggregate-list, function-list)

The 'group' function aggregates the input table on the key list.
For each partition, it applies each element of the function list to
the corresponding column in the aggregate list, creating a new field
whose name is formed by joining the column-name to the function-name.
For example,

	group(employee, age, salary, avg)

returns a table with key age and aggregate field avg_salary, containing
the average salary for employees of the same age.  The 'group' function
should support at least the aggregation functions count, sum, and average.

The 'group' function provides us with SQL's 'group by':

	select avg(salary) from employee group by age

The syntax of use will depend on the implementation language.  The 
goal of the exercise is to produce a tool-set which is most useful
to the application programmer, not the end-user of the application.

Optional additional functions
-----------------------------

5)	table <- insert(table, rows)

6)	table <- update(table, column-list, row-list, rows)

7)	table <- delete(table, selection-criterion)

What is tested: Databaselike data storage