Re: Inconsistence in transaction isolation docs

Albe Laurenz ha scritto:
> Nico Sabbi wrote:
>
>> /From:
>> http://www.postgresql.org/docs/8.2/interactive/transaction-iso.html
>>
>> "
>> Read Committed/ [...]
>>
>> to me the above sentence sounds inconsistent: it's
>> asserting that both 1) and 2) apply:
>>
>> 1) it never sees ... changes committed during query
>> execution by concurrent transactions
>>
>> 2) Notice that two successive SELECT commands can see
>> different data, even though they are within a single
>> transaction, if other transactions commit changes
>> during execution of the first SELECT
>>
>> Can anyone explain, please?
>>
>
> 1) means: as long as the first SELECT runs ("during
> query execution"), you won't see changes made by
> another transaction.
>
> 2) means: when you run a second SELECT, that SELECT
> will see changes made by other transactions, even if
> both SELECTs are in one (read commited) transaction.
>
> That doesn't sound contradictory to me.
> There is a difference between "during query execution"
> and "within a single transaction", maybe that is where
> your problem comes from.
>
> Yours,
> Laurenz Albe
>
>

well, I know how read committed behaves, but
I don't see why should anyone expect an update of the resultset
of the currently executing query after a commit by a different
transaction.
Thanks everybody who replied,
    Nico