cols t vs key flip value t

At least twice in tick.q

key flip value t is used to get the columns names of the table t.

Would cols t not do the same job?

If so, surely key flip value t has a greater memory overhead.

Perhaps the author was trading memory for speed?

 

q)t:flip($'.Q.a)!26 10000000#26000000?1000 q)(cols t)~key flip value t 1b q)\ts:10000 cols t 12 528 q)\ts:10000 key flip value `t 2 880

One of the loveliest aspects of working in q is how easy it is to experiment like this!