Suppose the stored query S (stored in U) is:

      <willbe name=a value="g_sum(city;;sales)" />

and suppose the input query Q on T is:

      <sel value="city='newark'" />
      <willbe name=a value="g_sum(city;;sales)" />

We say that Q matches S because Q matches the evaluation of:

      <willbe name=a value="g_sum(city;;sales)" />
      <sel value="city='newark'" />

so the query R

      <sel value="city='newark'" />

can be evaluated on U.

This is so because g_sum on break-column city is indifferent to whether the selection occurs before or after the g_sum.

Now consider the following example:

Table T has columns state and city, and no city belongs to more than one state.  The stored query S is as above, and the input query Q is:

      <sel value="state='new jersey'" />
      <willbe name=a value="g_sum(city;;sales)" canselect='state' />

The canselect attribute lists those columns whose extensions are supersets of the extensions of the break columns.  We can therefore know that the evaluation of Q will match the evaluation of:

      <willbe name=a value="g_sum(city;;sales)" canselect='state' />
      <sel value="state='new jersey'" />

and therefore that we may replace Q on T with R on U:

      <sel value="state='new jersey'" />