nuke pls

morning

@PaulVargas Check the room logs before that (and SCIENCE" ;)

@MartinSmith @PaulWhite is my above comment correct? ^^

I don't like ot be spreading hoaxes and I'm not entirely sure that this is correct and related to the question.

Oh, I found an answer by Remus that confirms

@ypercubeᵀᴹ ^^^ I think their screenshot demonstrates that.

@AndriyM OK. Converted the index scan to a gazillion index seeks - if I understand the plan correctly.

Their demonstration uses an index hint.

@PaulWhite you around?

I'd like to excuse myself for yesterfday

I have a tendency to be irked really quickly by people thinking dead people are something to joke about

I shouldn't I know

but still, thanks for saving the chat yesterday by bringing up a random topic like F1

@ypercubeᵀᴹ It's true but not especially relevant there. NEWID in that case is only evaluated once. So when seeked, there is a single seek.

@TomV Yes hi.

@TomV Absolutely no worries. I apologise if I was too harsh on you. The sudden development caught me rather by surprise.

@PaulWhite I'l lremove my comments then

@ypercubeᵀᴹ It's a really interesting discussion though. Consider: the same plan with a seek versus the same predicate in a filter. How many times is newid() evaluated in each case?

The general answer is that it is not guaranteed either way, and in particular the optimizer provides no guarantees about the timing or number of evaluations of a scalar function. These things are best avoided completely by making it explicit (perhaps by putting the desired value of newid in a variable first).

In case that's not clear: seeking to the value obtained by calling newid once is not the same as filtering each row after a scan if we call newid for each row.

Yes, this pattern of not guaranteed / unexpected behaviour exists in other DBMS as well (mysql, Postgres), with random() or similar functions.

It's a fundamentally tricky thing.

Making things like GETDATE runtime constants can be seen as an attempt to resolve that ambiguity for a common class of query in a way that is likely to surprise people the least.

I've been following some related discussions regarding current_timestamp and similar functions in Postgres lists.

Related to replication and parallel plans

@ypercubeᵀᴹ Do they handle it differently?

They introduced parallel plans only in the latest version (9.6) so one discussion was about some obscure bug that was introduced.

and how to handle it and avoid it in general when they add more options to the planner for parallel plans.

I find the way combinations of features can conspire to produce edge case behaviours fascinating.

@PaulWhite I'm not sure about the specifics. Postgres has some options on how functions are declared (immutable/stable, parallel safe/unsafe, etc.)

That's ok. I was only mildly curious.

Winter Bash 2016 countdown has started

seconds?

@PaulWhite I can't really explain, but wether it be a terrorist attack or a few officers killed, I dislike joking about dead people

I wonder how much time will pass before someone complains at meta that the countdown doesn't have days, hours, minutes

And I think in here those are the only 2 reasons I really got irked by someone

all other jokes are acceptable

Depending on the length of explanation.

You weren't too harsh, I was

I would have even accepted a kick mute because honestly I needed to calm down

but the change of topic was a much better solution

We so rarely get any problematic behaviour in here. Hardly any of us are used to dealing with it. Jack Douglas has a natural talent.

@PaulWhite Yes, but you handled it well, I misbehaved and I apologize

And if you accept my apologies, that 'll be the end of this episode

not only you, the heap, I messed up

@TomV All good with me.

@PaulWhite Thanks

Never mind, I had to make an edit anyway.

@TomV good man. Not sure I should tell you this but I think Room Owners can unban themselves?

@JackDouglas opening Pandora's (or someone else's) box?

@JackDouglas Are you suggesting we validate that assertion for science?

@PaulWhite Sorry, was away. Thnx for the edit. (That's what happens when we delete comments! ;)

@JackDouglas too late ;)

@JackDouglas @PaulWhite even if the ban (kick-mute whatever) is from amod?

This one can be closed as duplicate now: dba.stackexchange.com/questions/156980/…

needs 4 more votes

> But my databases do not look healthy. dba.stackexchange.com/questions/157100/…

@ypercubeᵀᴹ they should start eating better. start with a gluten free diet