On 2017-03-01 11:18:34 -0500, Tom Lane wrote:
> Andres Freund <andres@anarazel.de> writes:
> > On 2016-12-08 13:34:41 -0800, Andres Freund wrote:
> >> Hi,
> >>
> >> I'm wondering if it's not time for $subject:
> >> - V0 causes confusion / weird crashes when PG_FUNCTION_INFO_V1 was
> >> forgotten
> >> - They have us keep weird hacks around just for the sake of testing V0
> >> - they actually cost performance, because we have to zero initialize Datums, even if
> >> the corresponding isnull marker is set.
> >> - they allow to call arbitrary functions pretty easily
> >>
> >> I don't see any reason to keep them around. If seriously doubt anybody
> >> is using them seriously in anything but error.
>
> I find these arguments pretty weak. In particular I don't buy your claim
> that we could stop zero-initializing Datums, because I think we'd just
> start getting valgrind complaints if we did.
I tink we've litigated that in a side-thread - I'm not planning to change
any policy around this. Perhaps I shouldn't have quoted the original
start of the thread...
At this point I primarily am concerned about
a) the way they're confusing for authors, by causing spurious crashes
b) at some point not too far away in the future I want to introduce a faster interface, and I don't want unnecessarily
manyaround.
Greetings,
Andres Freund