[Nitro] [PATCH] glue-validations-to-og

Fabian Buch fabian at oggu.de
Sat Apr 15 02:52:27 EDT 2006


Am 15.04.2006 um 01:04 schrieb Bryan Soto:
> On 4/13/06, Dimitri Aivaliotis <aglarond at gmail.com> wrote:
>>   1) Failure:
>> test_all(TC_OgAggrCalc) 
>> [./test/og/tc_aggregations_calculations.rb:41]:
>> <28> expected but was
>> <58.0>.

> This one is the test. The other databases sort the result set.
> Postgres doesn't. The consensus on the list was to not assume the
> results were sorted. I need to do that...

I doubt that you can rely on other stores sorting it if you don't 
specify ORDER BY. If ORDER BY is not given and it is sorted, then you 
had good luck. Some databases sort in the order you entered the data, 
some other by the position on the harddrive and other things.

For Postgres:
http://www.postgresql.org/docs/8.1/interactive/queries-order.html
"If sorting is not chosen, the rows will be returned in an unspecified 
order. The actual order in that case will depend on the scan and join 
plan types and the order on disk, but it must not be relied on. A 
particular output ordering can only be guaranteed if the sort step is 
explicitly chosen."

How should it know how to sort your data? (not everyone assumes "id" or 
"oid" is the right field to sort by, only Og always creates an "oid" 
field)

I'm not so familiar with MySQL and not at all with SQLight, but I'm 
very sure the same I said for PostgreSQL above applies for Oracle (not 
supported by Og yet, so no issue here).
My guess is, that even for MySQL it was luck that it was sorted for the 
testcase.

Fabian




More information about the Nitro-general mailing list