From: | "David E(dot) Wheeler" <david(at)kineticode(dot)com> |
---|---|
To: | Steve Crawford <scrawford(at)pinpointresearch(dot)com> |
Cc: | PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: generate_series() Interpretation |
Date: | 2011-06-27 18:38:32 |
Message-ID: | AA76EC8B-7880-466F-A614-8A6DF4060E29@kineticode.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Jun 27, 2011, at 11:36 AM, Steve Crawford wrote:
> The query is marginally trickier. But the better calendaring apps give a variety of options when selecting "repeat": A user who selects June 30, 2011 and wants a monthly repeat might want:
>
> 30th of every month - skip months without a 30th
> 30th of every month - move to end-of-month if 30th doesn't exist
> Last day of every month
> Last Thursday of every month
>
> Typical payday repeats are "the 15th and last -day-of-month if a workday or the closest preceding workday if not", "second and last Friday", "every other Friday"...
>
> No matter how '1 month' is interpreted in generate_series, the application programmer will still need to write the queries required to handle whatever calendar-repeat features are deemed necessary.
Yeah, which is why I said it was subject to interpretation. Of course there's no way to tell generate_series() which to use, which is what I figured.
Thanks,
David
From | Date | Subject | |
---|---|---|---|
Next Message | Robert Haas | 2011-06-27 18:50:45 | Re: Range Types, constructors, and the type system |
Previous Message | Steve Crawford | 2011-06-27 18:36:14 | Re: generate_series() Interpretation |