kdb+ conditional insert: only insert when column value doesn't exist - insert-update

Question

What is the best way to insert a row into a table, only if a column of that row doesn't exist in the table.
E.g.:
q)table:([] col1:(); col2:(); col3:());
q)`table insert (1;2;3);
q)conditionalInsert:{if[first where table.col1=x(0)~0N;`table insert x]};
Now when doing the following:
q)conditionalInsert[(1;2;3)];
q)conditionalInsert[(7;8;9)];
The result yields:
q)table
col1 col2 col3
--------------
1 2 3
7 8 9
This can probably be accomplished more easily. My question: what is the easiest/best way?
To be clear: the column may be a non-keyed one.
Or in other words: Table is either keyed or non keyed and target column is not a key (or part of the compound key columns)

Answer 1

Use a keyed table?
q)table
col1| col2 col3
----| ---------
1 | 2 3
q)`table insert (1;2;4)
'insert
q)`table insert (2;2;4)
,1
q)table
col1| col2 col3
----| ---------
1 | 2 3
2 | 2 4
you can always use protected evaluation to silent the error.
q).[insert;(`table;(1;2;4));{`already_there}]
`already_there
q).[insert;(`table;(3;2;4));{`already_there}]
,2
q)table
col1| col2 col3
----| ---------
1 | 2 3
2 | 2 4
3 | 2 4

Answer 2

First thing is to have proper attributes (sort,group) on the target column which will make function faster.
Now there are 2 scenarios I can think of:
a) Table is keyed and target column is keyed column : In this case normal insert will work in way like your conditional insert.
b) Table is either keyed or non keyed and target column is not a key (or part of the compound key columns) :
q)conditionalInsert: {if[not x[0] in table.col1;`table insert x]}
Its better to use 'exec' in place of 'table.col1' as dot notation doesn't work for keyed table:
q)conditionalInsert: {if[not x[0] in exec col1 from table;`table insert x]}