Re: [GENERAL] Question about conccurrency control and Insert

2003-09-11 Thread Stéphane Cazeaux
It's now clearer for me. Thanks a lot for you long explanation. -- Stéphane Richard Huxton wrote: On Wednesday 10 September 2003 12:54, Stéphane Cazeaux wrote: Richard Huxton wrote: Client2's first SELECT started before you commited the INSERT, the second SELECT

Re: [GENERAL] Question about conccurrency control and Insert

2003-09-10 Thread Stéphane Cazeaux
Richard Huxton wrote: On Wednesday 10 September 2003 08:34, Stéphane Cazeaux wrote: Client 1: BEGIN; SELECT count FROM test_count FOR UPDATE; --> returns the only entry "1" Client 2 : BEGIN; SELECT count FROM test_count FOR UPDATE; --> this query is blocked, ok

Re: [GENERAL] Question about conccurrency control and Insert

2003-09-10 Thread Richard Huxton
On Wednesday 10 September 2003 12:54, Stéphane Cazeaux wrote: > Richard Huxton wrote: > > > >Client2's first SELECT started before you commited the INSERT, the second > >SELECT started after you commited. Since you are using READ COMMITTED you > > can read the results of transactions committed *bef

Re: [GENERAL] Question about conccurrency control and Insert

2003-09-10 Thread Tom Lane
=?ISO-8859-1?Q?St=E9phane_Cazeaux?= <[EMAIL PROTECTED]> writes: > I'm ok about this, but, if I try exactly the same scenario, where I > replace the INSERT by a DELETE, the first SELECT of the client 2 won't > return any row. This is the same behaviour with an UPDATE. If client 1 > updates the ro