2008/12/28 Tom Lane :
> I've spent quite a bit of time reviewing the window functions patch,
> and I think it is now ready to commit, other than the documentation
> (which I've not looked at yet at all). Attached is my current patch
> against HEAD, sans documentation. This incorporates the recent
"David Rowley" writes:
> Hitoshi Harada wrote:
>> I tested on Oracle 10.2.0, and the results are:
>> ...
>> which means the section 4.15 is true. Could anyone try DB2?
> DB2 9.5 results [ are the same ]
OK, good, that means the reference to the frame in 6.10 rule 1b is just
a copy-and-pasteo. (
Hitoshi Harada wrote:
> I tested on Oracle 10.2.0, and the results are:
>
> select depname, empno, salary,
> lead(salary, 1) over (order by salary),
> lag(salary, 1) over (order by salary),
> first_value(salary) over (order by salary),
> last_value(salary) over (order by salary)
> from empsalary;
Hitoshi Harada wrote:
> I tested on Oracle 10.2.0, and the results are:
>
> select depname, empno, salary,
> lead(salary, 1) over (order by salary),
> lag(salary, 1) over (order by salary),
> first_value(salary) over (order by salary),
> last_value(salary) over (order by salary)
> from empsalary;
> -Original Message-
> From: Robert Haas [mailto:robertmh...@gmail.com]
> I looked at this some more. I'm a little concerned about the way
> we're maintaining the in-memory hash table. Since the highest legal
> statistics target is now 10,000, it's possible that we could have two
> orders
Why does pqGetHomeDirectory have to succeed to use conn->sslrootcert.
Maybe this should be an OR of the two since sslrootcert is not dependent
on homedir?
around line 970 src/interfaces/libpq/fe-secure.c
if (conn->sslrootcert || pqGetHomeDirectory(homedir, sizeof(homedir)))
--
Andrew Chernow
Hitoshi Harada wrote:
> 2008/12/28 Tom Lane :
>> "Hitoshi Harada" writes:
>>> 2008/12/27 Tom Lane :
which doesn't conform to spec AFAICS ...
>>> 4.15...says:
>> interesting...6.10 general rule 1b, which very clearly states ...
>> ... 4.15 does seem like evidence that the spec authors may
2008/12/28 Tom Lane :
> "Hitoshi Harada" writes:
>> 2008/12/27 Tom Lane :
>>> I notice that the current patch code seems to implement
>>> first/last/nth_value using the frame, but lead/lag using the partition,
>>> which doesn't conform to spec AFAICS ...
>
>> In 4.15, it says:
>
>> The lead and la
"Hitoshi Harada" writes:
> 2008/12/27 Tom Lane :
>> I notice that the current patch code seems to implement
>> first/last/nth_value using the frame, but lead/lag using the partition,
>> which doesn't conform to spec AFAICS ...
> In 4.15, it says:
> The lead and lag functions each take three argu
2008/12/27 Tom Lane :
> "Robert Haas" writes:
>> Unfortunately, if we don't want to add an explicit iswindowable flag
>> (and I understand that that's ugly), then I think this is the way to
>> go. It's a shame that people will have to make code changes, but
>> inventing a fake AggState object jus
2008/12/27 Tom Lane :
> I notice that the current patch code seems to implement
> first/last/nth_value using the frame, but lead/lag using the partition,
> which doesn't conform to spec AFAICS ... but lead/lag on the frame
> doesn't actually appear to be a useful definition so I'd rather go
> with
2008/12/27 Tom Lane :
> The last bit of performance-related hacking that seems to be needed in
> the window functions patch is to fix things so that we can trim old
> rows from the underlying tuplestore when they're no longer needed.
> In particular I think it's critical to be able to do this for t
Hi,
Attached is the self-contained patch to skip checkpoint at pg_start_backup.
This is a part of Synch Rep patches, and was discussed in the following
thread.
http://archives.postgresql.org/message-id/3f0b79eb0812240710j7e613f3atfd6b6fc274035...@mail.gmail.com
In Synch Rep, we basically have to
13 matches
Mail list logo
