Tom Lane wrote:
> Joe Conway <mail@joeconway.com> writes:
>
>>I was thinking something similar. This exact question has come up at
>>least three times in the last three months. I doubt we'd want a special
>>keyword like CURRENT_QUERY, but maybe current_query()?
>
> Not unless you want to promote a quick debugging hack, not expected or
> required to work 100%, into a supported feature. I don't think
> debug_query_string can be relied on to always reflect what the system
> is doing, particularly not in the 3.0 protocol extended-query case.
> And how about when you're executing queries inside a function --- is it
> supposed to tell you about the most closely nested SQL query?
>
> I don't say this is not worth doing --- but I do say you are opening a
> larger can of worms than you probably think.
>
Hmmm. Good points. This one may best wait for 7.5 at least. Does it make
sense to turn it into a TODO?
* promote debug_query_string into a documented, supported feature
Anyone who *does* use the function from dblink, please be sure to report
circumstances where dblink_current_query() returns something other than
what you would expect.
Thanks,
Joe