On 10/7/2008 9:02 AM, Andrew Sullivan wrote:
> On Mon, Oct 06, 2008 at 06:05:30PM -0400, Christopher Browne wrote:
>> There is also a fourth possibility, which is to decide, at the time of
>> subscription, by copying some data into some sort of "test table" in
>> BINARY mode, and then pick whether or not to use BINARY based on
>> whether the test worked.  That avoids requiring users to fiddle with
>> parameters altogether.
> 
> That sounds like a bit of a hack.  But what about storing, in a Slony
> table at set-up, information about the architecture &c?  There must be
> some way to get this, although it could be more or less expensive.  If
> it's not available, you put "unknown" in, and provide a tool to get it
> and update it later.  I'm thinking of something that reads the output
> of pg_config or something of that sort.

The devil is usually in the details. Even though you might be running on 
the same hardware, how would you figure out automatically that some user 
defined data type stores slightly different binary datums depending on 
the version of some 3rd party toolkit?

No matter what we do, we still need the ability to override a false 
positive. So why not start with a manual switch on SUBSCRIBE SET and see 
how much performance gain that gives us?


Jan

-- 
Anyone who trades liberty for security deserves neither
liberty nor security. -- Benjamin Franklin