On 02/22/2018 05:23 PM, Ken Tanzer wrote:
On Thu, Feb 22, 2018 at 5:05 PM, Adrian Klaver
<adrian.kla...@aklaver.com <mailto:adrian.kla...@aklaver.com>> wrote:
On 02/22/2018 04:58 PM, Ken Tanzer wrote:
On Thu, Feb 22, 2018 at 4:53 PM, Adrian Klaver
On 02/22/2018 04:44 PM, Ken Tanzer wrote:
Hi, hoping to get some help with this. I'm needing to
specific date, a series of dateranges and, given a
date, return a single conitinuous daterange that
includes that date.
To elaborate a bit, I've got lots of tables that
and end dates. For example:
CREATE TABLE tbl_staff_assign (
staff_assign_id SERIAL PRIMARY KEY,
client_id INTEGER NOT NULL REFERENCES
staff_id INTEGER REFERENCES
staff_assign_type_code VARCHAR(10) NOT NULL
staff_assign_date DATE NOT NULL,
So a client might leave a progrma and then return
later, or they
might simply switch to another staff_id. (In which
record will have and end date, and the next record will
the next day.) In this case I need to know "what
they continuously in the program that includes X date?"
like to be able to do something like:
) FROM staff_assign sa2 WHERE
I've done this before with procedures specific to a
table, and working with the start and end dates. I'm now
wanting to try to do this once generically that will
all my cases. So I'm hoping to do this in a way that
performance isn't horrible. And it's a little unclear
to me how
much and how I might be able to use the daterange
accomplish this efficiently.
The operator I use to solve similar problems:
@> contains element
@> '2011-01-10'::timestamp t
Thanks Adrian. But how would you apply that to this situation,
where I have a series of (quite possibly discontinuous) dateranges?
This is going to depend on a more formal definition of the problem
with some sample data. Right now I am trying to reconcile "what
period were they continuously in the program that includes X date?"
with "where I have a series of (quite possibly discontinuous)
dateranges? " Maybe its just me, I can't see how discontinuous can
also be continuously.
-- Episode 1
-- Episode 2
Given 1/15/18, should return 1/1/18-->NULL
I have similar case. There I use the coalesce to substitute the date the
report is run for the NULL value, mainly because I have not way of
predicting the future:)
Is there actually an episode value you can sort by or is that a made up
value for this example?
Given 1/15/16, should return 1/1/16-->4/30/16
Thinking about this a bit more, I'm wondering if a window function could
be used for this? I've used them a bit, but definitely never wrote one.
OVER (PARTION BY client_id)
But then a window function can only see the rows included in the query,
yes? In which case this would only work if you queried for all the rows
for a particular client_id? I guess in the example I gave I was doing
that anyway, so maybe this would be no worse.
A Free Software data system
By and for non-profits
Subscribe to the mailing list
learn more about AGENCY or
follow the discussion.