It's rumoured that efesar once said:
>
> For now I will make the SQL box read-only. I know there are other
> solutions, but are too costly to implement right now.
Hi Keith,
Some development on that. We're discussing some changes to the wire
protocol for v7.4 on the pgsql-hackers list, and I've requested the
addition of the attrelid and attnum columns to the RowDescription method
whereever possible. This means we will know exactly where each attribute
in a resultset has come from, so we can easily find out any info about it.
If the message leaves those values blank, then we know there is noi
underlying base attribute, so we cannot update (eg and aggregate, function
call or undeterminable column from a natural join).
I think it's going in, I've answered all the questions that Tom Lane has
posed, and there are a few other strong supporters of the proposal.
>> Anyway, now it does some things here, it looks excellent - Nice work!!
>
> Thanks. The next patch will add tons of functionality. At this point it
> can create 85% of the queries you'd ever build.
Excellent, looking forward to it.
Regards, Dave.
