Re: Inspect values of prepared statements

Поиск
Список
Период
Сортировка
От Tom Lane
Тема Re: Inspect values of prepared statements
Дата
Msg-id 14231.1462225259@sss.pgh.pa.us
обсуждение исходный текст
Ответ на Re: Inspect values of prepared statements  (Natalie Wenz <nataliewenz@ebureau.com>)
Ответы Re: Inspect values of prepared statements  (bricklen <bricklen@gmail.com>)
Список pgsql-admin
Natalie Wenz <nataliewenz@ebureau.com> writes:
> So, I tried turning statement logging on and that didn't give me any more detail. I think that is because the query I
wantto inspect is generating an error (which I failed to mention in my original question). 

> ERROR:  invalid input syntax for type double precision: "A"
> STATEMENT:  insert into x (a,b,c,d,e,f) VALUES ($1,$2,$3,$4,$5,$6)

> columns c, d, e are all double precision, so I'm not sure which one has the bad data. If I knew what values $1 and $2
were,I could pinpoint the record that is generating the error. 

Ah.  You're out of luck on that, I'm afraid, because what the
parameter-value logging code does is to print from the converted internal
forms of the parameters.  If you get an error while converting the values
to internal form, which is evidently what happened here, no log for you.

Also, that conversion happens on-the-fly as each Bind message field is
read.  Thus, for example, if the bad input is for parameter $2 then we
simply can't print the remaining parameter values, because we have not
seen them yet.  (This could be fixed, perhaps, but it would involve a
lot of restructuring and probably slow things down.  Also it wouldn't
work for parameters arriving in binary form; which is probably what led
us to make the choice to do parameter logging this way.)

What I think we could do without too much trouble or overhead is to set
up errcontext data so that a failure like that could come with a CONTEXT
line like

ERROR:  invalid input syntax for type double precision: "A"
STATEMENT:  insert into x (a,b,c,d,e,f) VALUES ($1,$2,$3,$4,$5,$6)
CONTEXT: reading parameter $2 for statement "foo"

I'm curious whether that would be enough information to solve your
problem.

            regards, tom lane


В списке pgsql-admin по дате отправления:

Предыдущее
От: "Joshua D. Drake"
Дата:
Сообщение: Re: Inspect values of prepared statements
Следующее
От: bricklen
Дата:
Сообщение: Re: Inspect values of prepared statements