Jan Wieck <JanWieck at Yahoo.com> writes:
> On 10/6/2008 1:54 PM, David Rees wrote:
>> On Mon, Oct 6, 2008 at 10:30 AM, Jan Wieck <JanWieck at yahoo.com> wrote:
>>> Any sort of "automagically" will usually lead to false negatives (which are
>>> annoying) as well as false positives (leading to errors). So even with an
>>> automatic mode, we would need an explicit override capability.
>>>
>>> I would suggest adding a BINARY=yes/no option to the SUBSCRIBE SET command.
>>> What needs to be compatible are the architecture and PG version of the data
>>> provider and the subscriber of the single SUBSCRIBE command itself, so it
>>> cannot be a global config parameter anyway.
>>
>> Did you mean to send this only to me, or just forgot to CC the list?
>> Anyway, your solution sounds good to me. :-)
>
> No, it was meant for general ... thanks for catching that.

Ah, yes, and that's a good catch.  Whether BINARY is OK to use does
depends on the path.

Actually, that suggests a third place to configure it, namely as part
of STORE PATH.  I'm not sure I *like* the idea of configuring it
there, but it's certainly plausible.

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.
-- 
let name="cbbrowne" and tld="linuxdatabases.info" in String.concat "@" [name;tld];;
http://www3.sympatico.ca/cbbrowne/languages.html
C program run -- Run program run -- Run, C program, Run! -- (please)