Re: [HACKERS] COPY with no WAL, in certain circumstances

"Simon Riggs" <simon(at)2ndquadrant(dot)com> writes:
> On Sun, 2007-01-07 at 03:53 -0500, Tom Lane wrote:
>> I think you just talked yourself out of getting this patch applied.

> Maybe; what would be your explanation?

The main reason is that you were guilty of false advertising. This
patch was described as being an application of a known-and-agreed-safe
optimization to a new case, viz letting COPY into a new table use a
whole-file fsync instead of WAL-logging individual records. I suspect
most people didn't look at it closely because it sounded like nothing
very new; I certainly didn't. Now we find out that you've also decided
you can subvert the MVCC system in the name of speed. This is NOT
something the hackers community has discussed and agreed to, and I for
one doubt that it's safe. The active-portal kluge that you've just
mentioned is nothing but a kluge, proving that you thought of some cases
where it would fail. But I doubt you thought of everything.

In any case the correct method for dealing with a new optimization of
questionable safety or value is to submit it as a separate patch, not
to hope that the committer will fail to notice that the patch doesn't
do what you said it did.

regards, tom lane