The Heap™ – Consultancy ©®

4:26 AM

@JamesLupolt been fiddling with 2389 and 2390

I'd stick to 2389

2390 has all kinds of "watch out, be careful!" attached to it

@AaronBertrand that's good info. what other than "dont use this without 2389" have you found?

@swasheck that it can go very wrong

i dont know how to respond other than to laugh

@swasheck If you call support they will likely tell you to turn it off and solve another way. What have you tried to resolve this issue before resorting to 2389/2390?

4:36 AM

@AaronBertrand delete

(I'll take a look in the morning, getting the stink-eye)

@AaronBertrand are you asking me or are you telling me what they'd say. because i dont use it in prod, it just looks like something interesting to fiddle with in a lab

@AaronBertrand bahahaha. later.

oh ok, I thought "fiddling" meant "solving problems"

nah

though i wouldnt put it past me

You will definitely see more frequent compiles, you may see longer compile times. Also the TF may be a no-op in 2014, at least under the new CE

4:42 AM

@AaronBertrand so far that's been the result of my fiddling. 2371 is something we are running in production and, so far, the tests with prod replicas are not yielding consistent/predictable results with either 2389 or 2389/2390

so 2371 has covered most of our use cases

Yeah 2371 is definitely more reliable. It's just too bad they haven't gone beyond 200 steps or made a more complex histogram for these scenarios. You could always switch to GUID and have blazing fast write times too...

though the new CE and the incremental stats have really changed our approach toward stats management. incremental stats is probably going to take away the utility of 2371 altogether

Yeah partition-specific was a long-time coming. How that wasn't in 2005 I have no idea.

Ok really in trouble now

cya

go away ... in the nicest way possible

later

:)

Andriy M

8:01 AM

Coming from this person, I think this reaction could qualify as "well", even if not "very well":

@AndriyM: Yes, you are right. The rules for comapring to null are different between C# and SQL. — Guffa 20 mins ago

The downvote wasn't mine but still there was one and that's his reaction. So there you are

@AaronBertrand: Quite the opposite. Not equal to is far more intuitive than less than or greater than, you are just more used to one way of writing it. Familiarity doesn't mean that it's intuitive. — Guffa 22 mins ago

muhaha, less than or greater than

(and that's about SO for today)

and a snowy good morning

11:27 AM

so Guffa is the new Aptem?

also, weird stuff:

I think the queries you posted don't work at all. The (SELECT "name","type","area","z_order") part should throw an error saying ERROR: column "name" does not exist. — dezso 36 secs ago

James Lupolt

Heard a new tactic from a recruiter today: "A client has asked me to contact you."

they must have been very content

12:59 PM

morning folks

bluefeet

morning

morning

in my comment above, I was wrong

I hope I'm not here (with emphasis on the SQL Server part):

Your IDs do not like random at all. To repeat myself, I'd expect a difference only in adding the new IDs to the index - page splits and reordering would happen more often than in a sequential PK. But this is what happens with an average non-PK index anyway, so it must not be that terrible. (In SQL Server, for example, when the table is clustered along the PK, the picture would be very different.) — dezso 51 secs ago

@dezso whatever happened to Aptem

nice comment, did not think about it

@JNK Has anything happened to Aptem?

1:11 PM

@ypercube haven't seen them around

@dezso I haven't dealt with it myself but I read some whitepapers. Someone (I forget who) showed that on upper end SSDs GUIDS perform better than IDENTITY CI keys because the random IO is so much faster.

It's definitely not something that is relevant for most cases but it bears keeping in mind :)

1:28 PM

@ypercube You mean he-who-must-not-be named?

Q: SQL Server 2012 Installation Has No Option To Uninstall

> The short version is their IT department has had some issues, I think the only person still working there is the guy that makes the tea

5

1

I've recently inherited a laptop for testing from a client, unfortunately it wasn't a fresh image so it has some software already installed that's causing me some grief. The short version is their IT department has had some issues, I think the only person still working there is the guy that make...

@dezso starred

1:57 PM

Drinkies next friday all

@ConcernedOfTunbridgeWells I'll get some beers for supporting you from home

@dezso That's the spirit.

@ConcernedOfTunbridgeWells the real spirit would be me flying to London :) this is more easily communicable, though

2:13 PM

@JNK I didn't quite find that. Even on SSDs GUIDs were slower, but the delta is diminishing. Modern hardware does start to wave the arguments away, but I'm still a minimalist - I will take the downsides of integers over the downsides of GUIDs, because I'm not building Kejser's "we need 8 bazillion inserts a second and we'll never read any of it" systems.

@AaronBertrand Plus, if you have >4B records you can still use bigints. Ain't nobody got 1.7*10^19 records in a real system.

@AaronBertrand that's who it was

Kejser

Change of topic, does anyone here use visual studio much?

@JNK On a semi-regular basis.

I'm just discovering studio shell and it's pretty amazing

@ConcernedOfTunbridgeWells the argument is that increasing integers lead to hotspots in high-insert environments, and the ascending key problem (especially if data growth isn't uniform over time).

@JNK I don't believe I have it installed on a single machine at present (except for the shell stuff that Management Studio requires).

I tried SSDT at one point but couldn't get into it.

2:19 PM

@JNK What are you doing with it?

@ConcernedOfTunbridgeWells not sure if you have to refactor/maintain large codebases but if you do you may wanna read about studioshell

It's a powershell provider for visual studio that treats all your solutions like drives

and lets you do mass analysis quickly

and or mass changes

All the files/classes/members are objects in a tree

So I can say "show me all the public boolean properties that don't begin with the string "Is" in the solution"

and then if I want I can mass update them

and all references to them

That could be quite useful. The VS extensions API is quite byzantine. Being able to write extensions or scripts through a more warm and fluffy powershell API is probably a big win.

it works with the debugger too

so you can say put a breakpoint everywhere in the code base that X is used

That sort of thing could be very useful.

I just heard about it on a podcast yesterday and am playing with it today

it's free/open on github

er codeplex

studioshell.codeplex.com

2:49 PM

Apparently the CIA's headquarters is called the 'George Bush Centre for Intelligence.'

And who says Merkins don't get irony.

Q: PostgreSQL SELECT primary key as serial or bigserial query review

3:04 PM

yesterdaily WTF

1

I've meshed together a way to determine what the data_type is as in the data_type you use in the syntax when creating a new table (not the political definition) based off of the PostgreSQL wiki page and my earlier attempts to finalize my internal documentation of minimal query-to-productivity cro...

postgresql primary-key datatypes sequence catalogs

John's comments are interesting, to be mild

Its going to take me hours to figure out how to create a stored procedure for this simple report. If someone could write it for me, I would be sincerely grateful. Email me at robert.brady AT metals4u.com — Robert Brady 5 mins ago

again...

I mean, hours, really?

well, starting from nowhere... you should write it in 5 minutes and bill him for those hours ;)

@dezso You would be welcome. Berlin is not far ;)

3:15 PM

@ypercube it is not far $$ when you buy your tickets in time :D

3:25 PM

@AaronBertrand good grief

@JNK how about reading some tutorials instead of throwing your hands in the air and saying "I don't know how to do that." Using big words like implement doesn't help your case, either.

:-)

Q: Error converting .DBF into SQL or Excel

@AaronBertrand throw your hands in the air, and wave 'em like ya just dont care

bluefeet

utter laziness

-5

I have already posted the question on SQL Central, so just copying the link here. If anyone has answers to this please answer here. https://ask.sqlservercentral.com/questions/117717/converting-dbf-file-into-excel-or-sql-server.html#answer-117727

ssis dbf

@bluefeet vote to reopen, just to vtc again

4:30 PM

@AaronBertrand they are hopeless

THis Worked OrderDate=LEFT(oRS("ordDate"),4) OrderDate=Replace(OrderDate,"/","") TodayDate=LEFT(Date(),4) GetToday=RIGHT(TodayDate,2) GetYesterday=RIGHT(Orderdate,2) IF GetToday - GetYesterday = 1 THEN — Robert Brady 39 mins ago

@Robert I think that is quite possibly the worst way to do that. Riding a tricycle to work "works" but it's going to take me all day. Other methods not only work, but they work better. — Aaron Bertrand ♦ 43 secs ago

4:53 PM

I just had the most obnoxious conversation of my professional career.

@Zane "what's 1 + 1? 3"

Long story short it was about how we have to fill a column with Unknown because Null means something else to the business user.

@Zane i would push back

John M

Could always fix at the presentation layer, no?

Then someone said that BI requires 11 different types of NULL. Now my brain hurts and I'm very angry.

4:56 PM

@Zane 11 different types of null? NuLL, nUlL, NULl ...

@Zane Unknown, Not applicable on this type of record, Supposed to be there but not recorded, Value in error, Late arriving data ...

I got into it a bit and then just rolled over and died.

Its datetime I just hope its a simple process, my (hunt it down, break it and hack it til it works) skills are basic. LOL — Robert Brady 21 hours ago

^whenever is see a dumb comment then LOL I picture the person saying it loud and dumb.

Like a cartoon character. Derp Derp I don't know what I'm doing LOOOOOOL

Shawn Melton

@JohnM your comment on this question regarding comma as a decimal (that I guess you deleted) is not confusing. Some language formats use a period as a thousand seperator and comma as the decimal. Just FYI.

Dear Robert
     Fuck you,

Sincerly everyone

Its going to take me hours to figure out how to create a stored procedure for this simple report. If someone could write it for me, I would be sincerely grateful. Email me at robert.brady AT metals4u.com — Robert Brady 2 hours ago

John M

@ShawnMelton hence the deletion :)

5:07 PM

Its going to take me hours to figure out how to create a stored procedure for this simple report. If someone could write it for me, I would be sincerely grateful. Email me at robert.brady AT metals4u.com — Robert Brady 2 hours ago

Hmmm it's not letting me do this.

Shawn Melton

@JohnM Ah, I did do a double take on it the first time I read through it.

John M

@ShawnMelton I'm used to seeing it in currency, can't recall last time I saw it just for a number though

James Lupolt

@JNK Which podcast was that?

5:23 PM

@RobertBrady We just need written approval from your employer. — Kermit 9 secs ago

@ConcernedOfTunbridgeWells Sounds fun.

You know there are 11 ways to unfold a cube in 2d: mathforum.org/library/drmath/view/54682.html

Assign each shape to each kind of null ;)

Oops, it should be to @Zane

@JamesLupolt PowerScripting #270

7:23 PM

@ypercube hahaha

@Zane I like visual representations ;)

Lamak

7:50 PM

Again, I'm not looking for a letter. I'm looking for a word that is one letter long. There is a huge difference between the two. — Jonathan Allen 36 secs ago

really?

Sure, "bob is funny" contains an f, but it's not a word that is one letter long. "Just f that guy" contains a word that is one letter long.

(I did not click so I have no idea about the context. Jonathan is pretty ornary - he explains a question in vague terms and then gets mad when people pry for details. I'm going to delete this.

Lamak

yeah, but a fulltext index won't help with that kind of search, right?

Why do you say that you wouldn't worry about duplication of code? This is outright messy and un DRY. — Pacerier 9 mins ago

@Lamak hmm, not sure

i forgot. who here works int he MSP region?

7:55 PM

Zane I think

Lamak

@AaronBertrand so, it might help with that kind of search?, interesting

@Lamak not sure, I know what full text is but I don't use it. I would suspect a one-letter word is noise, but I don't know if there are ways to fool FT into not doing so