Jon Nelson <jnelson+pgsql(at)jamponi(dot)net> writes:
> Initially, I mis-typed the query, and one of the columns specified in
> the query doesn't exist, however the query ran nonetheless.
> The actual query:
> select gid from t2, t3 where t2.name = t3.name and t3.scope = 'city'
> and t3.hierarchy = 'STANDARD' and t2.adiv = t3.adiv limit 1 ;
> However, there *is* no column 'name' in table 't2'.
This is the old automatic-cast-from-record-to-text-string issue,
ie it treats this like "(t2.*)::name".
We've been over this a few times before, but it's not clear that
we can make this throw an error without introducing unpleasant
asymmetry into the casting behavior, as in you couldn't get the
cast when you did want it.
BTW this seems pretty far off-topic for pgsql-performance.
regards, tom lane
In response to
pgsql-performance by date
|Next:||From: Robert Haas||Date: 2010-10-29 18:13:31|
|Subject: Fwd: failure notice|
|Previous:||From: Kevin Grittner||Date: 2010-10-29 17:18:20|
|Subject: Re: MVCC and Implications for (Near) Real-Time Application|
pgsql-bugs by date
|Next:||From: Tom Lane||Date: 2010-10-29 17:56:23|
|Subject: Re: What happened to SSL_CIPHERS? |
|Previous:||From: Magnus Hagander||Date: 2010-10-29 17:39:04|
|Subject: Re: What happened to SSL_CIPHERS?|