« first day (2479 days earlier)      last day (2389 days later) » 
00:00 - 23:0023:00 - 00:00

12:01 AM
Mobile
 
@sp_BlitzErik Want to see my new row goal demo?
 
Eating popcorn
Sure
Is it an answer to that should be closed top 500 question?
 
yes
why should that be closed?
there are two answers and no close votes
 
No plans, no view definition
 
> needing plans to answer questions
I get what you're saying
but the way he worded his question made it answerable
it's just not a good question
I'll always have sympathy for row goal questions
row goals are why I created an account here
 
12:08 AM
It's half rant
 
haha
if you want to see a rant check out my first question
2
Q: Clustered index scan appears to be costed too low with row count spool

Joe ObbishI've never posted one of these before so let me know if any information is missing. I have a relatively simple query that performs pretty poorly with the default sample size on a few key columns. The costing of a clustered index scan used by a row count spool appears to be impossibly low. I will...

 
> This is the SQL Server version that I'm testing on:
Microsoft SQL Server 2014
Smh
 
what
 
12:30 AM
Basically 2012
 
it was over a year ago
 
1:02 AM
@sp_BlitzErik Already added to my answer
But thank you
"Produce content with long-term value" -- well, there go all my answers. — sp_BlitzErik 13 hours ago
Ha!
 
I was just happy that I finally found a question that Erik already hadn't answered
 
@JoeObbish Now we've established that challenges aren't off-topic, are you going to try setting one? Or ask a new meta about how they should be presented? I don't mind either way.
 
@PaulWhite Did we establish that? Maybe I missed something
 
@PaulWhite you're a peach.
 
@JoeObbish That's how I assess the outcome of the meta (going only by upvotes), yes. There are some concerns about topicality, but nothing that should prevent a good example being well-received I reckon.
 
1:08 AM
is there a way that I could ask the question so that upvotes go to me and downvotes go to you?
 
@JoeObbish It's pretty rare (here) to get overwhelming support on meta. As I read it, you're good to go, as long as you take the criticisms on board.
@JoeObbish I doubt you'd get much in the way of downvotes, but I've been wrong before.
Today, my only real reservation is keeping them within Q & A format, and preventing them taking over the site.
It does feel a bit like "fun" but I can get past that.
 
hmm
I did find it hard to express what I was after in the meta question
 
That's often the way.
Asking questions is hard anyway, probably doubly so on meta.
 
@PaulWhite I could make them not fun for you.
 
the only thing that comes to mind is "Can anyone construct a data set that gets worse compression with columnstore than rowstore?"
I don't think that would be covered by the meta question
 
1:12 AM
@sp_BlitzErik Appreciated :)
 
or "Can anyone construct a demo where nesting views leads to a measurable performance hit compared to using derived tables?"
that one has bugged me for years
it's one of those things that seems like someone said it 15 years ago and everyone has repeated it since
 
Don't derived tables not get inlined the same way views do?
 
@sp_BlitzErik that's the thing
 
Like since they can't be referenced multiple times or whatever
 
I've never heard a serious warning along the lines of "don't nested too many derived tables"
In fact I go deep all the time
 
1:16 AM
Ah
 
all
the
time
 
Well, you have a blog
 
There's precious little difference between chaining CTEs, nesting derived tables, or views.
 
I KNEW IT
 
Nesting views is just something people do a lot.
 
1:17 AM
@sp_BlitzErik How is that relevant? I'm not following
 
@JoeObbish 🤔
 
@JoeObbish I think you'd have to think about that a bit and phrase it carefully to avoid getting trivial answers. It would need to be clear what success looks like as well.
 
right, as you said it's not an excuse to ask bad questions
I mean, it's not like chat or anything like that
 
I'm hoping snapshot views will help with some of the top-1/min/max/per group views.
 
has MS said anything publicly about them?
 
1:20 AM
@JoeObbish Not that I've seen.
 
@JoeObbish just that they're not compatible with column store.
Jk pal
 
Ooo you had me
 
@sp_BlitzErik no EAGL
you had me too
 
I was just about to ask for a link
 
ok I think I got a decent "challenge" question
do either of you know of a better source than sys.messages for getting random words?
within SQL Server, of course
 
1:23 AM
Nope. I dislike sys.messages quite a lot though. Last time I used a public download of the complete works of Shakespeare I think.
Just personal preference, wouldn't read too much into it.
Ha ha read
 
too late
way better than Shakespeare imo
 
@JoeObbish What am I looking at there?
 
"to JOIN use generated re-establishing scan memory-optimized"
beautiful
a random string generated from sys.messages
 
Oh OK.
 
"Failed has not to function."
can't argue with that
 
1:32 AM
@PaulWhite what's your issue with it?
 
@sp_BlitzErik plan
 
I was at first confused and then amused by the halloween protection needed when inserting from it into another table
 
@PaulWhite serial? Underlying query?
 
ugliness
sorry eating my lunch
 
I got it
I want to be one of those people who uploads plans of DMV queries to paste the plan
it's disgusting!
although Paul may be referring to something else
 
1:38 AM
@PaulWhite do enjoy
@JoeObbish I think it's cute
 
new yorkers have such poor taste
oh yeah
I need to go buy lemonade apples
but before that, Erik I have a dumb hardware-related question for you
 
Shoot
@JoeObbish we can afford it.
 
@sp_BlitzErik I think it's when it's joined to itself a few times. I seem to recall thinking the plan was particularly ugly; and yes, the serial thing.
Compared with the standard constant scan numbers generating thing, anyway.
I guess beauty is very much in the eye of the beholder.
Maybe it wasn't sys.messages at all. Ho hum.
 
1:56 AM
Three cross joins with it will slow things down iirc
@JoeObbish email me the question, going to bed soon
 
will ask later
@PaulWhite The store is sold out of lemonade apples
Can you send me one through the Post?
 
@JoeObbish That's your punishment for procrastination!
Let's hope you have learned an important lesson today
 
hold on here
I had to do research before such an important purchase
why should I be punished for that?
 
2:20 AM
Because life is unfair
And competition for apples is strong
 
Will you send me an apple if I stump you with a sql challenge question?
 
Probably not
We can't just be sending apples willy-nilly
Also: export restrictions and biosecurity
I could probably send you a picture of an apple
 
in the Post?
 
Fax if you like
waits for Joe to look up what a fax is
 
sticking with the MS Access theme, eh?
 
2:26 AM
Very much so
I've decided to fully embrace the retro
The modern world is just too confusing
 
first step is to ditch the smartphone
 
I mean, storing data in columns sheesh
My phone barely qualifies
NZD 99 Huawei something or other
Oh it's a Huawei Y6 Elite apparently. Would hate to see the non-Elite model.
 
I knew you were an elitist
 
Ooo! Review Queues are Red! AWOOGA
@JoeObbish A proper phone
I had a Motorola Star-Tac once
The Motorola StarTAC is a clamshell feature phone manufactured by Motorola. It was released on January 3, 1996, being the first ever clamshell/flip mobile phone. The StarTAC is the successor of the MicroTAC, a semi-clamshell design that had been launched in 1989. Whereas the MicroTAC's shell folded down from below the keypad, the StarTAC folded up from above the display. In 2005, PC World put StarTAC at #6 in The 50 Greatest Gadgets of the Past 50 Years. The StarTAC was among the first mobile phones to gain widespread consumer adoption; approximately 60 million StarTACs were sold. == Release... ==
 
first ever, eh?
 
2:32 AM
According to WP
 
I was 7 when that came out
 
I wanted a microTac but was out of my price range at the time
@JoeObbish In the modern world you'd be on your third or fourth phone by that age
 
I can't understand that at all
 
Me either but there we are
 
3:25 AM
One more week of identicon
 
3:39 AM
3
Q: How does the review-needed indicator work exactly?

Jon EricsonWith the rollout of the new top bar, you’ve probably noticed the new review indicator: It’s possible there will be a different review icon by the time you read this. But this post is just about the red circle on the icon. You’ll see the indicator is turned on: if any review queue has a red d...

Right at the moment, there is a bug that prevents the indicator from lightning up for people with less than the moderator tools privilege. (See this screenshot.) The bug should be fixed soonish. — Jon Ericson ♦ 9 mins ago
 
3:51 AM
"Hi there,I check your blog named 'Create a website or blog at WordPress.com' daily."
the benefits of improperly configuring your site
 
 
2 hours later…
6:11 AM
@JoeObbish Yeah
Morning
 
7:16 AM
Morning
 
Morning
 
Morning
 
@PaulWhite It always makes me chuckle when you say that
 
@TomV I've adjusted to it :)
Never mind it's 8:20pm
 
@PaulWhite you live in Australia?
9
 
7:19 AM
@AndyK Don't you start.
6
New Zealand.
 
@PaulWhite lips sealed, Sir ;))
 
That's more like the level of respect I should get around here!
See if you can educate the others :-/
 
@PaulWhite will try my best (or something like that)
 
Good on ya
 
We're incorrigible
 
gbn
7:21 AM
He's clearly not Australian, he said "morning" not "G'day cobber"
 
going back to mining my dirty data
 
7:38 AM
@TomV The Heapizens shall never be corridged!
 
8:00 AM
Oh, Heapsters.
 
 
1 hour later…
9:22 AM
@ypercubeᵀᴹ thinking about our conversation yesterday
I found something similar today
SELECT dbo_T02_Customers_Kunshan.[customer code] AS Customer_code,
       Nz([Update_name],[customer name]) AS Customer_name,
       IIf([dbo_T02_Customers_Kunshan].[Global customer]="azo","azo",IIf(Nz([dbo_T02_Customers_Kunshan].[Global customer],"")="","#ND",[dbo_T02_Customers_Kunshan].[Global customer])) AS Global_customer,
       dbo_T02_Customers_Kunshan.Address1,
       dbo_T02_Customers_Kunshan.Address2,
       dbo_T02_Customers_Kunshan.Address3,
       dbo_T02_Customers_Kunshan.Address4,
Would this be the same as this query
SELECT dbo_T02_Customers_Kunshan.[customer code] AS Customer_code,
       Nz([Update_name],[customer name]) AS Customer_name,
       IIf([dbo_T02_Customers_Kunshan].[Global customer]="azo","azo",IIf(Nz([dbo_T02_Customers_Kunshan].[Global customer],"")="","#ND",[dbo_T02_Customers_Kunshan].[Global customer])) AS Global_customer,
       dbo_T02_Customers_Kunshan.Address1,
       dbo_T02_Customers_Kunshan.Address2,
       dbo_T02_Customers_Kunshan.Address3,
       dbo_T02_Customers_Kunshan.Address4,
 
@AndyK No, not necessarily. The EXISTS would be equivalent to IN (or to INNER join if tmp_nameunified_customer_azo.Customer_code is the PK)
Not to LEFT JOIN.
They would be equivalent only if there is an FK that makes the LEFT join equivalent to INNER join.
or if the query has extra WHERE clause that makes the LEFT join equivalent to INNER
 
@ypercubeᵀᴹ ok
@ypercubeᵀᴹ The LEFT JOINin the first query is here to make sure that if there is a match for ` dbo_T02_Customers_Kunshan.[customer code]` in tmp_nameunified_customer_azo.Customer_code then take the row.
too bad there is no other way to do that
 
@AndyK What is the tmp_nameunified_customer_azo table needed for? How does it participate in the SELECT clause? Is it for the Update_name column?
 
9:38 AM
@AndriyM thanks for that. you're correct.
it is for Update_namecolumn
this can be dropped then
 
If it's only one column that you are pulling from that table, than one other way is to use a correlated query.
If Access supports correlated queries, I mean
And I'm assuming here that tmp_nameunified_customer_azo.Customer_code has no duplicate values.
 
If we happen to meet one day, I treat all of you but one by one to the restaurant
@AndriyM no duplicate values, yes
@AndriyM thanks for spotting that
 
Please keep in mind that correlated queries are often not very efficient (probably less efficient than equivalent left joins, on average).
I only mention this method because it does present an alternative to a left join, and sometimes it works well enough, just not very often.
 
@AndriyM there is not much use of a correlated query in here. Most of the queries she has written were to amend datas that she could not ask the team in China or elsewhere to amend because she got embroiled with them.
the table tmp_nameunified_customer_azo is not much of use here, it contains only 6 lines
 
10:00 AM
I suddenly turned into a how-to-set-up-a-printer hero
the joys of moving to a new office
 
morning
 
 
5 hours later…
3:23 PM
see ya
 
@PaulWhite My dad used to have an old klaxon. It did, indeed go awoogah if you powered it with a six volt battery.
 
 
1 hour later…
4:41 PM
@sp_BlitzErik Looking good thanks
 
5:35 PM
@dezso be very careful
 
5:46 PM
@TomV surprised you're not all over this
1
Q: Troubleshooting non-working Plan Guide in SQL Server 2008 R2

user8217802I have problems with a Plan Guide I have created in Microsoft SQL Server 2008 R2. The Plan guide is purposed for a performance issue in Microsoft Dynamics AX 2009 and forces a specific question to use a TABLE HINT index. However, when I perform a SQL Profiler Trace, the execution plan still uses ...

 
6:03 PM
@JackDouglas any chance to add Oracle 12c in dbfiddle?
 
@ypercubeᵀᴹ last time he said there was a licensing issue
 
@sp_BlitzErik travelling today and in a restaurant right now
 
@JoeObbish Ah, thnx
 
which is definitely too bad
 
But thanks I hadn't seen it yet
 
6:06 PM
over four years old at this point
 
@JoeObbish you are?
 
He probably shouldn't be bossing the optimizer like that though
 
@AndriyM an edit, a bit ironic
(jeagl: AndriyM correcting identitiy to identity)
3
 
Ahh, I see
 
slow day, huh
 
6:11 PM
or night
both actually
 
@sp_BlitzErik well, yes
not sure what you were going for there
 
did you still have a hardware question?
 
yes
 
Interjection: your mom
  1. (pejorative, US, Britain, slang) A general purpose insult, often used to comedic effect due to becoming stale.
  2. Conrad: I am not sure if this phrase is commonly used...
  3. Dominic: Your mom is commonly used.
  4. 2004, Napoleon Dynamite
  5. Deb: Well, isn't anyone else here trying to earn money for college?
  6. Kip: Your mom goes to college.
 
you too
 
6:14 PM
Also are you sure the code you say that is working actually is working? You seem to have a single quote issue there — Tom V 1 min ago
Or am I missing something
I have Tom's phone not TomTom's so it won't spin up a vm
 
@sp_BlitzErik are you ready?
 
my soul is prepared
 
 
3 hours later…
9:26 PM
@sp_BlitzErik everything is burning to the ground
that is why I have not asked my question yet
 
@JoeObbish anything i can help with?
:D
 
I can pay you in popcorn
 
deal!
 
I don't understand...
 
looks like an fbi sting
 
9:36 PM
What in the name of small fluffy bunnies is gourmet popcorn
 
ask the new yorker
he can explain it at great length
 
it has strippers and cocaine on it
maybe a lobster
 
I was going to guess that you used champagne instead of oil to pop it
 
you eat it in a bathtub while wearing a top hat
2
note to self: paul has a notification for popcorn
 
9:54 PM
comments always make me regret answering
1
Q: Sql Server 2016 not using index used in 2012

SaxmanWhat do I check? I have almost identical tables with a slight difference in row count. One on 2012, and the other on 2016. Indexing is identical. This drop dead simple query for single record using a single column. 2016 version has 8254356 rows 2012 version has 8254427 rows

 
@sp_BlitzErik yep, also asking questions
 
at least i've never answered a question about access
 
me either
 
pinned messages disagree
but i suppose you'll pull some lamak-esque semantics to escape your fate
as a lowlife access dba
 
it was a chat discussion. there was not a question asked on the main site
also, your hero recently answered an access question on the main site
 
10:03 PM
i didn't know karl lagerfeld had a dba.se account
 
You'd be surprised who we have here hidden behind userxxxxxx names
And bland identicons
 
when brent fires me for spending all my time in this chat i'm gonna change my name to user9223372036854775807
 
shouldn't you change it before you get fired?
 
then he'd know who i am
cartoon is a dead giveaway
 
go for the identicon
 
10:11 PM
you're a worse criminal than a dba
 
all the cool kids are doing it
 
my identicon color is an awful blue
ain't happenin
 
don't be such a fashionista
Paul's is brown
it's the worst identicon I've ever seen tbh
 
@PaulWhite cheeky.
real cheeky.
 
@JoeObbish Six days to go.
> This does not give me confidence in the new optimizer.
This does not give me confidence in the OP.
@sp_BlitzErik What have I done now? :)
 
10:20 PM
mod stuff
 
Ah well that will happen.
 
until the rev.
 
Making the most of the time I have left
 
did you get your $100 in the mail?
 
@sp_BlitzErik Did you not see my reply?
But yes I did and it is fully awesome
 
10:23 PM
hah, no. i have a man servant that rings church bells when email from you shows up.
perhaps you didn't reply all.
 
@PaulWhite I suppose you can change it in 5 ;)
 
@ypercubeᵀᴹ Trust me I check it every day.
 
I mean you can say that it's NZ time
 
@sp_BlitzErik Just checked it came from Brent alone with no cc:
@ypercubeᵀᴹ Oh I see. Pretty sure Joe would be onto me pretty sharpish if I cheat.
 
Or change it today. SET TIMEZONE = +120 hours
 
10:28 PM
Still what's he gonna do.
 
Oct 10 at 22:46, by Joe Obbish
DATEADD(DAY, -2, THE_REV)
 
@ypercubeᵀᴹ I'd be lying if I said I wasn't tempted.
It had to be bloody brown eh
 
We'll find a question to distract him
 
Maybe I could send him an apple instead
 
tell him his row goal demo is wrong
 
10:30 PM
I try not to state the obvious
 
get the aloe
 
> I wont insert any records only import more records later on from a excel sheet
:wat:
0
Q: Is there anything I can do to increase performance in the future once table gets bigger?

anatp_123I created a table importing a excel sheet with 600,000 rows. Most of the queries I will run will be using C_CustomerID example: select * from testtable where C_CustomerID = 12345678 CREATE TABLE [dbo].[testtable]( [id_card] [int] IDENTITY(1,1) NOT NULL, [C_CustomerID] [int] NOT NULL, ...

 
BTW @sp_BlitzErik dba.meta.stackexchange.com/a/82 so please add as well as any version-specific tag.
 
didn't i?
 
10:36 PM
@sp_BlitzErik The previous question, the faulty optimizer one.
Regarding the latest one though: stackoverflow.com/q/44687182
Seems this is an ongoing struggle
 
oof
wait so the question i answered should have 2012 AND 2016 tags or just 2016?
i'm guessing just 2016
 
I would say so yes.
 
why can't i add tags to my answers?
"ace-answer"
 
Self-answered.
 
"voting-ring-activate"
"suck-it-joe"
 
10:38 PM
Ha!
 
oh boy, that's enough of trying to keep up with that
this is someone getting level 100 confused with level 400
 
@sp_BlitzErik Is it fun being able to see deleted questions now?
 
only when i'm going to answer them :D
 
> What the??? I think 2016 optimizer has been smoking something seriously strong.
grrr
 
yeah, that's where i bailed
I DON'T GET IT I'M ON A TOTALLY DIFFERENT SERVER WITH DIFFERENT SOFTWARE AND THINGS ARE DIFFERENT
 
10:46 PM
It is weird.
 
quite the oddity
i'd file a connect item
 
Time to go DEFCON 1 I reckon
Possible alien invasion
 
possible illuminati interference
they used the undocumented INTERCEPT operator to change the query plan
 
@sp_BlitzErik compare my answer to Brent's
I didn't want to say anything because I'm a polite guy
 
link
 
10:52 PM
14
A: 'SELECT TOP' performance question

Brent OzarSQL Server builds different execution plans for TOP 100, using a different sort algorithm. Sometimes it's faster, sometimes it's slower. For simpler examples of it, read How Much Can One Row Change A Query Plan? Part 1 and Part 2. For in-depth technical details, plus an example of where the TOP...

 
what should i compare? they're different questions
 
@sp_BlitzErik you're one of the people who voted them as duplicates!
 
yeah, i thought it should have been closed. voted to that end.
 
59 mins ago, by sp_BlitzErik
but i suppose you'll pull some lamak-esque semantics to escape your fate
 
00:00 - 23:0023:00 - 00:00

« first day (2479 days earlier)      last day (2389 days later) »