Re: Infinite loop for generate_series with timestamp arguments

I wrote:
> We could perhaps do better by doing the initial addition of the
> interval and seeing if that produces a value greater than, less
> than, or equal to the start timestamp. But I'm afraid that
> doesn't move the goalposts very far, because as this example
> shows, we might get different results in different months.

> Another idea is to check, after doing each addition, to make
> sure that the timestamp actually advanced in the expected
> direction. But should we error out if not, or just stop?

Here's a very POC-y patch using both of these ideas (and
choosing to error out if the interval changes sign).
If we go this way, generate_series_timestamptz would need
similar changes, and some regression test cases would be
appropriate. I'm not sure if the documentation needs
adjustment; it doesn't talk about the difficulty of
identifying the sign of an interval.

regards, tom lane