postgres column alias problem

In PostgreSQL you can not use expression with an alias in order by. Only plain aliases work there. Your query should look like this:

   select distinct           l2.*,           l.user_id as l_user_id,           l.geopoint_id as l_geopoint_id      from locations l left join locations l2 on l.geopoint_id = l2.geopoint_id     where l.user_id = 8  order by l2.geopoint_id, l.user_id = l2.user_id desc;

I assume you mean that l2.user_id=l.user_id ought to go first.

This is relevant message on PostgreSQL-general mailing list. The following is in the documentation of ORDER BY clause:

Each expression can be the name or ordinal number of an output column (SELECT list item), or it can be an arbitrary expression formed from input-column values.

So no aliases when expression used.

I ran into this same problem using functions from fuzzystrmatch - particularly the levenshtein function. I needed to both sort by the string distance, and filter results by the string distance. I was originally trying:

SELECT thing.*, levenshtein(thing.name, '%s') AS dist FROM thing WHERE dist < character_length(thing.name)/2 ORDER BY dist

But, of course, I got the error "column"dist" does not exist" from the WHERE clause. I tried this and it worked:

SELECT thing.*, (levenshtein(thing.name, '%s')) AS dist FROM thing ORDER BY dist

But I needed to have that qualification in the WHERE clause. Someone else in this question said that the WHERE clause is evaluated before ORDER BY, thus the column was non-existent when it evaluated the WHERE clause. Going by that advice, I figured out that a nested SELECT statement does the trick:

SELECT * FROM (SELECT thing.*,      (levenshtein(thing.name, '%s')) AS dist      FROM thing      ORDER BY dist) items WHERE dist < (character_length(items.name)/2)

Note that the "items" table alias is required and the dist column alias is accessible in the outer SELECT because it's unique in the statement. It's a little bit funky and I'm surprised that it has to be this way in PG - but it doesn't seem to take a performance hit so I'm satisfied.

You have:

order by l2.geopoint_id, l_user_id = l2.user_id desc

in your query. That's illegal syntax. Remove the = l2.user_id part (move it to where if that's one of the join conditions) and it should work.

Update Below select (with = l2.user_id removed) should work just fine. I've tested it (with different table / column names, obviously) on Postgres 8.3

select distinct        l2.*,        l.user_id as l_user_id,        l.geopoint_id as l_geopoint_id   from locations l   left join locations l2 on l.geopoint_id = l2.geopoint_id  where l.user_id = 8  order by l2.geopoint_id, l_user_id desc