« first day (2400 days earlier)      last day (2461 days later) » 
02:00 - 13:0013:00 - 21:00

2:09 AM
@PaulWhite Do you happen to know if dbcc csindex still works in SQL Server 2016?
 
@sp_BlitzErik $26M will get you one: theguardian.com/world/live/2016/mar/24/…
 
2:41 AM
@JoeObbish Yes it does
 
@PaulWhite Been stuck for a while. Do you happen to have a simple example that shows it working?
Can't figure out what I'm doing wrong here...
here's an example that doesn't work: pastebin.com/Qa0aqkKQ
 
3:00 AM
 
ok, there's the problem
column_id = 2
 
Yes
 
1 + the column_id from the DMV?
 
Half the fun with undocumented things is working them out for yourself
Remember there's a column locator / bookmark
Also, undocumented things aren't guaranteed to make sense, work consistently, or at all
DBCC CSINDEX is one, among many, that doesn't appear to have been updated as implementation details have changed underneath.
Like DBCC OPTIMiZER_WHATIF for example
 
ok, that works for me now
thanks
I've been trying to find all of the restrictions around aggregate pushdown
 
3:05 AM
And yes, I do understand that my definition of "fun" can vary from other peoples'.
@JoeObbish The extended event is a reasonable place to start with that
 
extended events aren't my style
 
Mine either. But there is one for that.
 
but I wonder how much easier that would have made it...
 
I can very rarely be bothered with XE
 
I'm totally with you on the "fun" part
except I've already spent so much time figuring this out
I can load the same data into columns in two tables
and AP works in one and not the other
 
3:07 AM
But sometimes XE is a useful thing to try to get started at least. The latest UI is not all that bad.
@JoeObbish I really would start with the XE. You might find the encoding is different, and not compatible with pushdown.
But yeah these things can be immensely time-consuming, sometimes for no reward.
That's part of the "fun".
 
I just didn't think it would be this hard
and I'm sure I'm missing stuff
do you already know all of the restrictions?
 
The Lehmer Random Number Generator did not come to me in a dream, for example.
@JoeObbish I'm sure I don't. But it's very hard to know the things one doesn't know.
 
good, there's a chance I can teach you something then
 
From a practical POV, agg PD either works in practice or it does not.
@JoeObbish Is that your goal?
 
No, my goal is to learn as much as I can for now
However, it would be nice to pay you back in some small way
 
3:11 AM
I'm sure a post that talked about things that prevent agg PD would be of interest to several people.
It's hard to judge in advance though.
I've always just gone with what interests me.
 
living the dream!
 
Sometimes that results in a post that has 100 lifetime views
Sometimes a few hundred thousand
I've never been good at assessing that in advance
 
do you try for SEO?
 
I do not.
That should be self-evident I reckon.
I believe Aaron does some work on that for the SQLPerformance.com articles
But it's all voodoo to me.
But hey you shouldn't necessarily listen to me on these things.
@sp_BlitzErik would know more if you're interested in volume
Or making money :)
 
deliberately didn't show ads
and no, that's not it
that's too much work
 
3:15 AM
Oh I didn't mean from ads, that's a horrible idea
I mean from promoting consulting/speaking/training services or whatever
So you could afford truffle coated popcorn, or whatever the BOU people have for breakfast these days
<\troll>
 
BOU?
and right, but ads are an easy way to get started at least
 
Brent Ozar Unlimited
 
could make my 2 cents per month
 
I hear Pinal Dave has a million dollar blog
 
wonder how much he costs to hire
I guess the goal is to have a reference for myself
doing any other goal properly requires SEO
 
3:20 AM
He actually submitted a session to PASS Summit one year with a title like that
Not sure if it got selected
Yes I started blogging specifically because I forget stuff if I don't write it down
Also, it consolidates knowledge, and makes one ask new questions
A bit like teaching others on a subject. You have to know it inside out to do that well.
 
I deleted pages of notes after you posted that stats answer
felt good
 
I nearly gave up on it actually
Was not straightforward
 
I ended up with a formula that worked sometimes
which is the worst
 
The end result looks simple enough, but then it always does
 
I had a table for which I figured out that auto sample used a repeatable value of 1
but others that didn't
so it would have been a while to figure that out, for sure
 
3:23 AM
Yes I went through a phase of plotting data points in Excel :)
Wolfram Alpha etc.
 
at some points I thought "it would really help to have a super computer"
still don't know how I missed the partition id...
feel like I got so close
 
I respect people that can make the brute force approach work, but it's not something I find intellectually satisfying
 
I had 2 CPUs to work with
 
You only need 1 CPU to step through the disassembly line by line :)
 
I only need 1 CPU + Paul White at my keyboard
 
3:27 AM
Optimizing compilers really are the bane of a reverse engineer's existence
 
ok I give up
I have what I have
I will perform a WAG
 
I spent ages trying to figure out what a block of instructions was doing, only to realize it was a fancy way of doing integer division/modulus.
@JoeObbish Sometimes writing the WAG down can help, even if you don't publish it
Try leaving it for 24 hours and coming back to it as well
I've had success with that
 
been at this on and off for like a week
good advice though
I don't mind writing up speculation
well, from a certain POV it's all speculation
I'll just run tests on SQL Server 2017
because I want to be one of the cool kids
 
@JoeObbish Without source code access, and/or access to the people that wrote the code, yes.
To a certain extent.
I find myself doing relatively little with pre-release software.
It's too easy to write stuff with a shelf life of 25 minutes.
There was a really interesting parallelism bug in one of the Denali RCs for example.
 
I'm just copying and pasting code
to see if any additional test cases pass
but yeah, totally agree otherwise
 
3:33 AM
I didn't write about it, because once I let them know about it, they fixed it in less than a week.
 
Niko has some stuff on agg pd but it's almost not useful
because sql server 2016 wasn't rtm yet
 
Sure.
 
interesting => bad?
 
Show stopper. They don't change RC code on a whim.
I believe code changes at that time require VP approval.
 
interesting
 
3:34 AM
For a brief time, yes.
 
we were watching parallel redo in 2016 pretty closely
I think it didn't make it into the final RC? if I remember correctly
we thought we were out of luck
 
That XE was query_execution_dynamic_push_down_statistics btw
 
I'm afraid to look
nothing in google?
 
Nope
 
ok
one improvement in 2017
that I can find
tbh I'm not sure why MS promotes this feature so much
maybe it sounds nice to people
 
3:49 AM
@JoeObbish What's that? Aggregate pushdown?
btw:
rows_not_pushed_down_due_to_pushdown_disabled
rows_not_pushed_down_due_to_encoding
rows_not_pushed_down_due_to_possible_overflow
(from the XE)
 
@PaulWhite yes
@PaulWhite :(
got some of those, missed others
 
I guess the first one is TF 9354. I haven't spent any time on this tbh.
@JoeObbish Do you not find it effective? Seems pretty awesome to me. Like pushing a Filter into a residual predicate for row store, but more so.
OTOH I'm not a huge fan of the way it is exposed in show plan.
 
Sunil's comment was pretty funny on that last point
 
Don't recall it. Where's that?
 
"Good point. At first, it indeed looks confusing. we have documented this and hopefully expert DBAs like yourself, who may need to look at this level of details, will know how to interpret"
 
3:55 AM
Oh yes that
 
my big issue with it is the encoding thing
 
They have some seriously bright people there, but goodness me they could do with some usability input sometimes
@JoeObbish Sure. That's not exactly transparent.
Maybe it will go away soon.
 
I still don't understand why rowgroup compression isn't independent
I'd ask but that's probably a long discussion
and it might be tucked away in the docs somewhere
 
Independent?
 
as in, the data that I'm loading into column B affects the size of the compressed segment for column A
 
3:57 AM
You're right that is a long discussion
 
no worries
overall I'm impressed by the number of things supported with agg pushdown
I tried some really stupid stuff
it's just the encoding
 
I always hoped they'd develop materialized views more
No matter how fast a bulk calculation is, you can't beat not doing it
 
do you agree that MS tends to develop new features for sql server and leave them in a "V1" state?
 
Nesting, outer join, apply, min/max etc.
I think that's a little cynical
They have limited resources like everyone
 
you don't find it fun to write the query in a way that avoids all of the restrictions?
 
4:01 AM
Not always possible, sadly
But yes
On the V1 thing, I think people need to understand the importance of feedback on the first releases. If they see little interest, it's unlikely to get future dev.
 
is interest different from production use?
 
It can be
 
what's the official channel for interest? connect?
 
For the masses, yes. Major customers have their own channels. It's a spectrum.
 
we recently got bumped up, I think
or maybe there was some kind of legal paperwork that finally got signed, not sure
 
4:05 AM
The sad thing about the mat.view stuff is that the theory is all there, and they've even built in into SQL Server builds, just never saw the light of day for whatever reason
 
how do you know? was it removed from RCs?
 
There are Microsoft Research papers on it
 
So not public builds
 
"sql server builds" includes internal only ones
 
4:06 AM
Yes. Where all the interesting TFs live :)
 
I've never gone through the private channel that we have
 
Well there you go :)
 
to be honest, I can't think of anything to ask that I don't already know the answer to
 
That must be nice
 
to clarify
often the answer is "we're not going to tell you that"
 
4:08 AM
Ha
 
if I was in their position I wouldn't tell me that either!
and if I did get an answer I'd have to keep it under NDA and partition it off somehow
 
Yeah well the point I'm making here is that people need to feed back to whatever contact they have (TAM etc) about V1 features.
@JoeObbish That's why the MVP thing can be a frustrating experience for some.
 
we've given lots of feedback on columnstore
 
And that's getting lots of love
 
yes, you're welcome
 
4:09 AM
Good on ya Joe
It'll be interesting to see if Graph gets any traction
 
sadly nearly all of 2017 isn't of interest to me
 
Same here
 
string_agg
 
But then 2017 ~= SQL Server on Linux
 
you can debug on linux!
 
4:11 AM
I'd have to learn a new OS
Not gonna happen
@JoeObbish Resumable online index build?
 
that's suspicious
 
How so?
 
I suspect I won't like how they implemented it under the hood
because I can't think of a way they could do it in which I would like it
for us we have long periods of the day with no writes
so not that interesting personally
 
is it enterprise only?
 
4:16 AM
No one knows until RTM
 
oh, right
sounds like it would be enterprise only
 
In the past, I would have said close to 100% likely, but the game changed
If I had to guess, I would say it would be all editions.
But I have no specific knowledge
 
care to wager on it?
 
If you can think of an interesting premise
Like if you lose you replace the default identicon with a recent face photo permanently
Not sure what I could offer
 
knew you'd go for that
you could replace yours with the default :)
 
4:21 AM
I am quite predicatable
Ok sure but let's limit it to, what ... a month?
 
hmmm
can I hold a sign in mine?
 
Like a sign with your identicon over your face? No!
 
it would not cover the face
 
Ok then
 
still deciding
I also won't take advantage of the loophole
(if I was to accept)
 
4:23 AM
Yeah I don't think we need lawyers
 
I'm a developer though
 
That's Ok I will accept the inevitable bugs in your implementation
Open to better ideas as well
Or a different time frame
 
not what I meant
as needed we can excel at meeting the literal requirements of a task
for example you just said "recent face photo"
I could use a picture of you to meet that requirement!
 
Clearly that would be a photo of you
 
you didn't specify
I'm struggling to think of the example
which makes this a really lame story
but I asked a room full of developers for an example of some kind of behavior in SQL Server
with an obvious one in mind
got a bunch of technically correct but useless answers
like "the table is empty"
 
4:27 AM
Yep I get it
 
Ok I accept your wager
 
Sweet
 
finished up my training on Friday
did the session on trace flags
 
Did you do the one about spools yet?
 
yes
I did them all!
that's what I meant
 
4:30 AM
Oh so it was the same talk?
I thought you were doing one on spools alone
 
two single hour sessions per week
for six weeks
 
OIC
Back later I have some chores
 
 
1 hour later…
5:32 AM
zǎo'ān (Chinese, Mandarin‌​)
 
5:52 AM
@PaulWhite One can only hope. That was my show-and-tell last week.
 
6:03 AM
@MichaelGreen Yes I think I saw a tweet from you about that?
How did it go?
 
6:57 AM
@sp_BlitzErik I got this just now with a filtered statistic.
 
morning
 
gbn
Morning
 
8:07 AM
Morning
 
8:33 AM
Morning
 
Morning
 
 
1 hour later…
9:48 AM
0
Q: change the state of database to online after mdf file deletion

SunilI am trying CorruptionChallenge3. Where the given solution recommends to delete all the file by switching database to offline mode. I have to change the state to ONLINE with ldf file. I am unable to alter state to online. Is it possible to alter state of database to online without .mdf file?

Can a mod delete this question PLEASE!!
No, seriously. What is the best reason to close this question? Off-topic?
 
unclear perhaps?
Ah, it's a competition.
Also requires downloading a corrupted database file.
@hot2use just flag it for mod attention and explain.
 
@ypercubeᵀᴹ That competition ended years ago, as of now it's just a practice
 
Page Not Found
This question was voluntarily removed by its author.
gone
 
10:13 AM
@TomV thnx. I didn't read very carefully.
It wasn't very clear what their issue was anyway
 
@ypercubeᵀᴹ It's an interesting series if you're on SQL Server actually
 
DBCC :-)
@TomV Yes, very. I followed the series when it was running.
 
10:27 AM
morning
 
 
1 hour later…
11:47 AM
@PaulWhite filtered statistics what now?
 
Jun 16 at 3:25, by sp_BlitzErik
*** WARNING: badly-formed histogram ***
^ The message I linked to.
Goodness me it's only just over a year ago what is wrong with you
 
Oh, heh. This chat thing is a real drag on mobile.
 
12:07 PM
There's a tiny little arrow at the start of the message even on mobile. Aiming for it with fat fingers is some of the most fun you can have on the Internet.
 
9 hours ago, by Paul White
And yes, I do understand that my definition of "fun" can vary from other peoples'.
 
not in this room
 
@ypercubeᵀᴹ I make a lot of sense
 
@PaulWhite I shouldn't be mixing the two uses though. I understood the mobile "fun" was sarcastic, while playing with TF and debuggers is real.
 
Everything's fair game here
Quoting out of context is fun
So we should be fine so long as bluefeet doesn't see it
 
12:19 PM
0
Q: Writing a simple SELECT Stored Procedure in Oracle PL/SQL

strangerThis is my SQL Server Stored Procedure which is below: CREATE PROCEDURE passenger_details AS BEGIN SELECT Full_Name, Age, Nationality, Category, Airline_Name, Class_Type FROM Passenger, Ticket, Airline, Class WHERE Passenger.Passenger_No=Ticket.Passenger_No AND Airline.Airline_No=Ticket.Airline_...

Really is no helping some people
 
@PaulWhite you've made an entire sport out of it
 
Not yet but it's gathering momentum I think
 
@Philᵀᴹ It's a hidden hint: "Please code for me"
 
That moment when you've checked in loads of code, merged it with a couple of other branches, then destroyed and recreated the dev VM.... then realise you forgot to git add a couple of new files ;(
 
@Philᵀᴹ Holey Moley
 
12:26 PM
Hence the expression, "you stupid git"?
 
I did utter some words, but they'd get me kicked out of this room if I repeated them :)
 
git is in heavy competition with powershell for being the worst
 
@PaulWhite "I didn't think Phil would Git it."
 
@hot2use 1154 commits in 6 months, so I do
 
@sp_BlitzErik I second that. Developed for C and C++ programmers and then open-sourced so that everybody has to use it.
And it's cool, cuz Linus developed it.
@Philᵀᴹ Ok. I surrender.
 
12:31 PM
@hot2use a lot of people think i'm just anti-CLI, but i'm fairly comfortable with them in other places (not just sqlcmd!)
 
I love the command-line
 
I assist my brain with a third-party tool. It's from the company that gave us bit-bucket
 
My favourite Unix command has to be cowsay
 
Moo! 🐮
4
 
ps -efx
 
12:33 PM
🐮
 
🐮🐮🐮
 
@sp_BlitzErik I don't know enough (or any really) powershell but what's wrong with git?
And aren't these two tools for different jobs?
Unless you are referring to learning curve
 
@ypercubeᵀᴹ answering the short one first: yes, but they're two tools that cause great pain every time i use them (and, well, you can run git commands in powershell)
the problem that i have with git is similar to the problem i have with powershell though. it's not terribly intuitive, documentation is fudgey, and the learning curve for jr/infrequent users is incredibly steep
2
 
git is fine. It's just when you have to do complicated merge branching & there are conflicts - can get a little counter-intuitive
GitHub is a godsend
 
@Philᵀᴹ Yes Sir.
 
12:39 PM
i couldn't live without that gui, yeah
 
@ypercubeᵀᴹ Did you read the Power Shell live blog?
 
@PaulWhite Yeah but I only skipped though it.
 
Mental image involving an actual skipping rope
2
 
@ypercubeᵀᴹ this is why you're the smartest guy in the room. it's really only worth skipping through.
 
Idea for blog title: Live Skipping
 
12:55 PM
if i ever have an office job again i'm going to skip rope at my desk to annoy people
 
Was doing some skipping in the gym the other week. It's surprisingly strenuous
 
i wonder if earnie shavers skipped rope
 
02:00 - 13:0013:00 - 21:00

« first day (2400 days earlier)      last day (2461 days later) »