Search notes:

SQL: Order of select operations

The order is:

`from`	to generate the data set
`on` / `join`
`where`	to filter interesting records from the data set
Add pseudo column `rownum`	Oracle only (See Select first n rows …)
`group by`	To specify which rows need to be combined for aggregations
`with cube` / `with rollup`
aggregation functions	to actually perform the calculations on the combined rows
`having`	To exclude (aggregated) records
Window (analytic) functions	to combine values from different records of the result set
`select`	Specify what is actually returned.
`distinct`	eliminate duplicates
Set operatiors (`union`, `intersect`, `except` etc.)	to combine results from multiple select statements
Model clause	Oracle only?
`order by`	to sort the result set
`offset`	to specify the first record that should be returned
`limit`, `fetch`, `top`	to specify how many records should be returned

Because aggregation functions are performed after group by, they cannot appear in group by expressions.

Because window functions are performed after aggregation functions, they can use the result of aggregation functions.

Because column aliases are defined in the select step, they cannot be used in preceding steps (such as the where clause etc.)

An example that combines order by and union all is here.

Links