Re: [HACKERS] log_statement output for protocol
От | Guillaume Smet |
---|---|
Тема | Re: [HACKERS] log_statement output for protocol |
Дата | |
Msg-id | 1d4e0c10608300025s2880837bi6fe3e7566b14edde@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: [HACKERS] log_statement output for protocol (Bruce Momjian <bruce@momjian.us>) |
Ответы |
Re: [HACKERS] log_statement output for protocol
|
Список | pgsql-patches |
On 8/29/06, Bruce Momjian <bruce@momjian.us> wrote: > Good point. I thought it was clear enough, but obviously not. I had a > similar case with bind, and used a comma to separate them: > > LOG: statement: prepare sel1, SELECT $1; > LOG: statement: bind sel1, $1 = 'a''b' For this one, I must admit I prefer the colon we used before. Something like: LOG: statement: prepare sel1: SELECT $1; LOG: statement: bind sel1: $1 = 'a''b' seems better to me as after the colon, we have the details of the command which is the common sense of a colon. > I am concerned a dash isn't clear enough, and a semicolon is confusing. > Using a comma the new output is: > > LOG: duration: 0.023 ms execute sel1 > DETAIL: prepare: SELECT $1;, bind: $1 = 'a''b' A dash is clearer in this case IMHO. ;, is not very readable. But I can parse easily both forms so it's not a problem for me if you prefer a comma. > Is that OK? Patch attached and committed. I also fixed the null bind > parameter bug. It now displays $1 = NULL (no quotes used). Cool. I'll test that. > Other suggestions? I'll compile this new version and make tests in the next few days to check everything is consistent and let you know. I'm still struggling to find a regexp good enough to parse "statement: execute y ('a', 4, 'text, with a comma'::text);" :). Thanks a lot for your work on this subject. It will help us a lot when we use the JDBC driver. -- Guillaume
В списке pgsql-patches по дате отправления: