| Subject | Re: [firebird-support] Re: How to implement SQLServer's WITH(NOLOCK) statment? |
|---|---|
| Author | Helen Borrie |
| Post date | 2008-04-14T14:17:36Z |
At 11:56 PM 14/04/2008, Martijn Tonies wrote:
However, as Thomas pointed out, even without the pessimistic lock, there *are conditions* where inappropriate settings on two read committed transactions can cause one of the transactions to be blocked from reading while the other has a write pending on the selected rows (or on the master row[s] where a FK relationship is involved).
./h
>> >> How about SELECT...FOR UPDATE WITH LOCKThat's why the documentation explicitly warns people to know and understand what they're doing when using the pessimistic lock. It is *certainly* not to be used with a query that returns more than a tiny few rows (ideally, one row).
>> >>
>> >> ?
>> >
>> > Right, I'll rephrase, "no plain select is ..." ;-)
>> >
>> > Either way, I don't think such a select would lock out other
>> > readers, but would only lock for other updaters?
>>
>> It will lock other readers as well. ;-)
>
>Hm, nasty.
>
>:-)
However, as Thomas pointed out, even without the pessimistic lock, there *are conditions* where inappropriate settings on two read committed transactions can cause one of the transactions to be blocked from reading while the other has a write pending on the selected rows (or on the master row[s] where a FK relationship is involved).
./h