Tom Lane wrote:
>Andrew Dunstan <andrew@dunslane.net> writes:
>
>
>>Do you expect turning it on to affect only future compilations? Or
>>should we recompile every function already compiled in the present
>>backend? I can see arguments either way.
>>
>>
>
>Yeah, me too. I would definitely expect a change in the variable
>(in either direction) to be respected in subsequent function
>compilations. I'm less excited about redoing previous compilations;
>a case could be made for doing so, but I won't insist on it.
>I think the main case where use_strict is interesting is in development,
>where you're repeatedly CREATE OR REPLACE'ing the function and retesting
>it, so you're going to be doing new compilations anyway.
>
>
>
>
Discussion seems to have died on this. If there's no objection and
nobody else is doing this I will prepare a patch based on keeping the
setting as USERSET and not recompiling previously compiled functions.
I'll move the embedding strings to macroland and tidy things up as
discussed. It will take me a few days to get that ready.
cheers
andrew