On Thu, 3 Aug 2000, Tom Lane wrote:
> Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> > I thought our only kludge was allowing = NULL because we got tons of MS
> > folks complaining in the past before we did this. Are you thinking that
> > the newer MS versions will not give us a problem.
> Well, I'm not sure. Magnus is saying that the newer MS servers default
> to spec-compliant semantics --- ie, foo = NULL will yield NULL. But
> IIRC the original complaints were because MS tools like Access would
> *generate* this expression and expect it to behave like foo IS NULL.
> Can MS have fixed all their apps already? Seems unlikely. Maybe we
> have to leave the kluge in there awhile longer.
I kind of agree with your original comment about removing the kludge,
since the kludge does go against the spec ...
... a couple of thoughts to that effect:
1. this won't be 'in effect' until v7.1 comes out anyway
2. v7.0.2 will still have the kludge and will be available if ppl *really*
need it, no?
My vote is to go with Tom on this and remove the kludge ...
In response to
pgsql-hackers by date
|Next:||From: Roland Roberts||Date: 2000-08-03 17:36:01|
|Subject: Re: Re: [HACKERS] random() function produces wrong range|
|Previous:||From: Tom Lane||Date: 2000-08-03 16:20:50|
|Subject: bit/varbit stuff|