Tom Lane wrote:
> Bruce Momjian <pgman@candle.pha.pa.us> writes:
> >> Nonsense --- a quick system() call in plperlu or pltclu can do it
> >> just as well, or even launch a shell script if you insist on coding
> >> your mistakes in sh...
>
> > Well, if they are going to call system() from perl, why don't we just
> > give them PL/sh and they can do it directly.
>
> There's a difference between making a dangerous thing possible and
> encouraging people to do it. But I'll wait and see what Peter thinks.
I think Peter agrees with you. My point is that we can PL/sh and
document it's limitations. With perl, the ability is there but it isn't
obviouis why it is a problem. Also, there must be some things you can
do in PL/sh that aren't transaction problems.
In my view, there are lots of people who need shell access from
PostgreSQL, and they are naturally going to try to use PL/sh. When they
do, we can explain the limitations. I don't see how not giving them
PL/sh prevents them from doing transaction-unsafe things.
Usually, people using PL/sh want to do something that is
transaction-unsafe by nature.
Also, there is a wiz-bang aspect to being able to do shell scripts in
the backend that has a certain "marketing" value.
-- Bruce Momjian | http://candle.pha.pa.us pgman@candle.pha.pa.us | (610)
853-3000+ If your life is a hard drive, | 830 Blythe Avenue + Christ can be your backup. | Drexel Hill,
Pennsylvania19026