From: | Andrew Dunstan <andrew(at)dunslane(dot)net> |
---|---|
To: | Postgresql Hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: selective statement logging |
Date: | 2004-03-10 22:36:51 |
Message-ID: | 404F9883.2000707@dunslane.net |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Bruce Momjian wrote:
>Andrew Dunstan wrote:
>
>
>>>Well, if it is parse error, then we can't know what type of command it
>>>really was. They could type 'SE9ECT 1' or 'SELECT 1 WH8RE x=1' and both
>>>are not really SELECT commands to me.
>>>
>>>
>>>
>>>
>>>
>>If I delay logging until after the parse, these log lines would come out
>>in reverse order. Not sure if that matters to anyone, but it might annoy
>>me slightly.
>>
>>line:3 LOG: statement: se4dt ddd;
>>line:4 ERROR: syntax error at or near "se4dt" at character 1
>>
>>
>
>Sure you sure? I didn't think you would get a tag on a syntax error, so
>no log would print, which I think is OK.
>
>
>
If people are happy with suppressing statement logging on a parse error,
OK. For the remainder I would just defer the logging till immediately
after the parse and get the tags for the statements - and suppress
logging the query string if they were all SELECT or all
SELECT|INSERT|UPDATE|DELETE|COPY according to the setting.
cheers
andrew
From | Date | Subject | |
---|---|---|---|
Next Message | Dennis Bjorklund | 2004-03-10 22:37:38 | Re: unsafe floats |
Previous Message | Neil Conway | 2004-03-10 22:36:35 | Re: unsafe floats |