Re: MySQL search query is not executing in Postgres DB - Mailing list pgsql-hackers

From Jeff Davis
Subject Re: MySQL search query is not executing in Postgres DB
Date
Msg-id 1354041921.10198.204.camel@jdavis-laptop
Whole thread Raw
In response to Re: MySQL search query is not executing in Postgres DB  (Robert Haas <robertmhaas@gmail.com>)
Responses Re: MySQL search query is not executing in Postgres DB  (Robert Haas <robertmhaas@gmail.com>)
List pgsql-hackers
On Tue, 2012-11-27 at 12:19 -0500, Robert Haas wrote:
> I admit that there are cases where this could happen, and that it will
> happen a little more than it does now.  But, as you say, this can
> happen now, and yet we get very few if any complaints about it,
> whereas we get regular complaints about the need to insert casts that
> other database systems do not require.  The fact is that most
> functions are not overloaded, so the esoterica of overloading affect
> only a tiny number of relatively sophisticated users.  The need for
> extra casts cuts a much broader swath through our user base.

Well, I did offer a suggestion that would make your idea safer, which is
to explicitly opt out of the overloading feature at the time the
function is created, rather than making it implicit based on how many
functions happen to have the same name.

The fact that it can only hurt sophisticated users is not convincing to
me. For one thing, our users are programmers, so they should all feel
comfortable defining their own functions, and I don't want to make them
any less so. Next, sophisticated users also make mistakes.

I could also make a security argument. Even today, any user who can
create a function in your search path can make your queries start
failing. If we locked down most of the system-defined functions as
non-overloadable, and allowed users to do the same for their functions
(maybe even the default one day?), then that would greatly reduce the
exposure.

The current strictness of the overloaded functions tends to make users
more explicit about argument types, which reduces the chance of problems
at the expense of usability and compatibility. Not ideal, but if we make
it more permissive then we are permanently stuck with less information
about what types the user intended and which function they intended to
call. In such an extensible system, that worries me on several fronts.

That being said, I'm not outright in opposition to the idea of making
improvements like this, I just think we should do so cautiously.

Regards,Jeff Davis




pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: Bugs in CREATE/DROP INDEX CONCURRENTLY
Next
From: Alvaro Herrera
Date:
Subject: Enabling frontend-only xlog "desc" routines