Bruce Momjian <pgman@candle.pha.pa.us> writes:
>> I certainly think it's worth considering, although I see Tom has
>> objected :-) . Say we have an option called FIRSTLINELABELS, then on
>> copy out it would write the headings on the first line, and on copy in
>> it could just ignore the first line (so it could work symmetrically).
> Yes, that is what I was thinking. Is this a TODO? Sure it can be
> misused but most copy options can be similarly misused.
Most copy options will result in obvious failures if misused. I dislike
this one because of the high risk of silent corruption of your data
(loss of a row or insertion of a row that shouldn't be there).
Considering no one has even made a positive case why we should add it
(for instance, a use-case where it's necessary), I don't think it
belongs in the to-do list.
regards, tom lane