« first day (2527 days earlier)      last day (2341 days later) » 

12:05 AM
@ypercubeᵀᴹ haha true
 
12:42 AM
5 people VtC this, yet there is only a single down-vote? It's a crap question, down-vote that sucka. It's free.
 
Thanks for the compliment, Max
not 100% mine, inspired by someone else
 
1:29 AM
It nicely captures the nuances of data distribution in lay terms.
 
1:44 AM
@ChrisTravers working on pg internals now
 
I'm an expert at communicating to the layman
 
@JoeObbish Never jumped into coding?
my goal is to teach PostgreSQL how to handle 512 bit hex encoded numbers.
using openssl bignum
 
 
6 hours later…
8:14 AM
evening and morning
 
8:34 AM
@EvanCarroll Nice! what sort of stuff?
:41545018
Nice :-)
 
8:52 AM
morning all
 
good morning / evening / midday
next
-1
Q: compare two columns and remove duplicates in the same same table

A.RI have table like id | t1 | t2 --------------- 1 | a | b 2 | c | a 3 | a | e 4 | f | g 5 | c | c I want to compare two columns with each other to remove matching records and get unique values in each columns, like id | t1 | t2 --------------- 1 | a | b 2 | c | e ...

 
9:19 AM
@hot2use The comments don't really clarify things :) I'm even more confused now than when I read the original question
 
@TomV it seems OP wants to maintain only first occurrence of t1 and t2
 
9:49 AM
@McNets I guess duplicate is the wrong phrase for that. They want a kind of first come first selected solution.
 
10:12 AM
@hot2use yep
In some way he needs to enumerate distinct occurrences of t1 and t2 in both columns and then null the corresponding values
 
@EvanCarroll If you need pointers, I can probably direct you to some things.
We have 32- and 64-bit base32 encoded numbers. in Pg.
No pun intended on the pointers bit.
 
10:30 AM
@ChrisTravers Was one intended on the bit bit? :)
 
@McNets Nice solution.
Sometimes I look at solutions and then realise that in some cases I am a real noob
 
@AndriyM Heh good one :-). Had not seen that. Interestingly one of the most interesting pointer problems I have seen is "the backend crashes when I run explain with my data type."
 
@JackDouglas We are still working on the ORA-600 issues. Next stop is to restart the physcial server as restarting the instance brought no relief.
 
The backend crash during explain but no other time is almost always caused by one specific error.
 
@ChrisTravers @ChrisTravers how hard would it be to create a VARINT type?
I mean a type that can hold arbitrary large integers?
And where should I start looking?
 
10:46 AM
@ypercubeᵀᴹ You mean like numeric?
 
but without floating points
So these are types represented by the varlena struct, and internally represented as an arbitrary string of base-1000 integers.
 
yeah. I don't know what structure, numeric type has
oh, thnx. You answered while I was asking
Why base-1000?
 
Because 2*10 is 1024, so you get fairly good alignment and overlap over time.
it means in 32-bits you get 2^30
sorry should be 2^10, not 2 * 10
the major limit is that arbitrarily large has limits due to TOAST limits
I think numeric is limited to something like 10^32000
 
So, in 64 bits, it would be 2^60 or rather 1000^6? and in 256 bits, 2^250 or 1000^25 ?
 
10:50 AM
sorry
you are right
hmmm that cannot be TOAST limits either but probably an addressing issue.
so maybe we could use bigger addresses and have a bigger ints.
 
I was thinking of a structure like VARCHAR
 
yeah, that is also a varlena type.
 
but base-2, without converting to base-1000
 
barchan is limited to 1GB of text.
*varchar
 
gbn
@ChrisTravers Is that like 4chan for SQL monkeys?
 
10:59 AM
hahaha
my typing today sucks
 
TypoCubeᵀᴹ is spreading the infection, it appears
 
@AndriyM 7 more days for him to pick a bash username
 
11:33 AM
@ChrisTravers The up key edits your last message
 
12:33 PM
Voting to close as unclear, until the OP provides us with the DBMS. — ypercubeᵀᴹ 1 hour ago
VTC please. Or add the SQL Server tag and reject the close. Seems the most likely bet.
 
@ypercubeᵀᴹ I would guess that's a screenshot from SSMS but I still VtC
OP should provide that information so we're sure
 
12:46 PM
@ypercubeᵀᴹ done
 
Hi. We are going to need a lot more information here. As it stands now I would answer the question with a simple 'yes there is a way to index the joins' — Tom V 33 mins ago
 
1:35 PM
@ypercubeᵀᴹ deleted by op after he edited it
 
2:32 PM
@AndriyM sir, thank you for the edit, sir ;)
 
Some developer teams are really dangerous...
0
Q: Is it dangerous to give production (read only) mysql credentials to the developer team?

me987654323All is in the title. Someone in my team want credentials to access a production MySQL database. Even in a "read only" mode I think it can be dangerous. What do you think?

 
gbn
2:48 PM
@McNets I did start a managing director level bun fight over a dev running SELECT statements on live
I saw it, told him to be careful and use ANSI-92 JOINs and use TOP etc.
 
"use nolock"
 
gbn
he instantly escalated to MD level. I won, he was fired soon after (for other reasons too)
 
Does anyone else remember a recent blog post from someone at Microsoft mentioning that they were working on some new language features including approximate query results, more adaptive query processing, and enhancements to string functions?
 
that was all in one post?
 
Yes, unless I'm imagining things. It was all for 2018 (or whatever vnext is going to be) I think
 
2:52 PM
i haven't seen anything official from ms
 
Thanks. It must have been something I read about about or heard from PASS.
i wouldn't mind finding a job that pays me to go to conferences again
 
@McNets I think giving a team a set of credentials is always dangerous. I think it is better to give users credentials.
 
@gbn We use a copy of our current db as a development db, obviously it is a quite small db
 
@JamesL did you mean thi slink, about SQL standard enhancements?
2 days ago, by Evan Carroll
@sp_BlitzErik https://s3.amazonaws.com/artifacts.opencypher.org/website/materials/SQLStandards‌​_2017-10-20.pdf
 
But now we are planning to buy SAP or SAGE X3 (at end) and we need to set up 3 db servers (production, test and development) due to FDA regulatory rules.
 
gbn
3:01 PM
@McNets This works for even bigger ones. I restore a 300GB one daily here, have done smaller and bigger
 
@ChrisTravers yep
 
gbn
The side benefit "restores are tested daily"
 
@gbn yes, you're right
 
hi tom
 
I bet the latter
 
3:06 PM
rough day?
 
The weeks after golive will be rough if they go through with that
The week starting with a SQL 2K issue was a bad sign
 
gbn
Are they deploying onto a SQL 2k box too...?
 
got nt 4 and a t1 line and everything
 
3:32 PM
@ypercubeᵀᴹ no i think it was something specific to sql server, probably from joe sack at pass. but i'll read that presentation, thanks.
 
 
2 hours later…
5:53 PM
@TomV lol, 9 year old bug closed bugzilla.mozilla.org/show_bug.cgi?id=474824
I love when those trackers tell you bugs open for a decade just got closed.
I got one from OO that was like 15 years old
I was wondering if it was my first FOSS bug
 
6:14 PM
I should map out the personal beef between famous FOSS developers, and whether or not it still exists. That would be hillarriuouusss.
 
6:44 PM
@EvanCarroll can you still repro? :)
 
7:00 PM
I'll be honest, I don't much use Firefox anymore.
=( but I can give it a shot when I pull down the 18.04 in another 4 months.
If you do webdev, FF isn't all that hot.
 
Quantum made me like it again
@EvanCarroll My personal favorite is this one which took 10 years to be assigned even if I included a 2 line patch a decade ago
 
7:20 PM
Looking back I find myself in an authors file but can't for the life of me remember what I contributed to that project :)
good times :)
 
7:41 PM
@TomV whatever it was, it was 10+ years ago.
 
Tom V (MUCH-APPRECIATED CONTRIBUTOR)
3
@JamesL Not sure what you mean by "enhancements to string functions". If you want info on the public upcoming adaptive query processing features I might be able to answer your questions
 
8:00 PM
there are a bunch of new string functions in 2017
but i haven't heard anything about 2018 yet
the only stuff i know about 2018 is nda
:'(
 
@sp_BlitzErik and they are quite useful IMO
 
yes, we're nearly on par with mysql now
 
well....yeah
but aside of the long overdue string_agg one, the translate and trim ones are nice
 
I don't know why we can't have GREATEST() and LEAST()
 
we can
when you learn c++ and get a job on the dev team
slacker
 
8:07 PM
@JoeObbish that would make things too easy
 
@sp_BlitzErik that's a task for an intern
so what did you think of that trace flag that I discovered?
did it change your life?
 
i was rocked to my very core
 
@sp_BlitzErik so you won't drink anymore?
 
yeah that's a reasonable conclusion
 
I've been working on being more reasonable, glad it's going alright
 
8:15 PM
I love it when a query takes twice as long when processing 1/10th of the data
 
sounds like parameter sniffing
ahem
 
there are no parameters in the query
 
outdated statistics?
 
oh
well the query has like a million joins
not looking for help
just complaining
 
@JoeObbish we can help you complain, though
 
8:19 PM
yes
somebody ask me if I can use temp tables
 
can you use temp tables?
 
@Lamak no, the application does not allow that
 
that sucks
damn devs
 
@JoeObbish good that you not mysql then.
 
why's that?
 
8:26 PM
@JoeObbish it's always good to not use mysql
no other reason needed
 
I didn't think about it like that
 
have you tried recompiling
 
don't be a one trick pony
 
@JoeObbish max of 61 tables joined allowed
 
@ypercubeᵀᴹ are you kidding? I wish there was a trace flag for that
that would force developers to rewrite their queries
 
8:52 PM
0
Q: What are the most relevant analogies between T-SQL and MDX?

EdmundI am proficient in MS T-SQL (CTEs, windowed functions, recursive CTEs, large data, table-valued functions, APPLY, sprocs, etc.) but have never queried an MS SQL cube. What are the top resources for a T-SQL user to expand into MDX on MS SQL Server? I am hoping there are some common analogies bet...

Still not sure about the question. He's still asking for resources and editing that out would make it borderline duplicate to the one I linked in my comment
I'll post a stub answer so I can follow up tomorrow but I won't cast a vote myself
 
9:22 PM
now this is weird:
0
Q: set postgres instance as readonly with no master

vinni_fI have a pg 10 instance that I want to behave as a read only/replica/standby instance. There is no master, only that read only database and I want the query select pg_is_in_recovery(); to return true to satisfy a monitoring check. I have tried add a recovery.conf file in the data dir with stand...

 
9:35 PM
@dezso or he could ignore the alert
 
10:10 PM
@ypercubeᵀᴹ no idea about the fake master port. This is something I don't even want to think about :D
if they figure out how not to log all the streaming errors, it might work
 
@dezso I just tried it, in 9.4 and 9.5
Yeah, I was about to edit my answer, that the log will be full of errors
@dezso First I only tried to stop a master and see if an existing alve will complain or not.
It didn't even after restarting the cluster
Then I took another cluster (not a slave, not a master), added the recovery.conf and restarted. Just works
 
10:52 PM
hello all
 
@JoeObbish I think it was about support for regex (or at least some type of pattern matching) in replace()
If it doesn't ring a bell no big deal
@swasheck hello
 
what do you all use for on-call escalation management?
 
a wiki page with a list of phone numbers
 
sweet.
 
i know that's not a very helpful response
 
10:59 PM
@swasheck pager duty - pagerduty.com
 
but i'm not sure what you mean i guess?
we use telegram for sending alerts to our phones
it integrates pretty well with scom, nagios, etc
 
@bluefeet cool. how do you get stuff to pagerduty? do you have some sort of APM?
@JamesL right. are you using SCOM? Nagios? Prometheus?
 
@swasheck scom for windows, nagios for unix and miscellaneous stuff, powershell and a gateway to telegram for most sql server related things
 
@swasheck number which then notifies the person oncall
 
@JamesL soooooooooooooo ... it's not unified :D
@bluefeet ah. didnt SO build their own monitoring service? it probably pushes to PagerDuty
 
11:04 PM
@swasheck no don't really see the benefit tbh. maybe i would if i was working with more people
 
where are you, these days?
 
an investment fund
london
 
nice.
 
i use bosun and opserver, just not for anything alert-related. pretty nice free tools
 
@swasheck well we use various alerts, etc that are seen by people which then could lead to escalating to the oncall person
we do use OpServer and Bosun to see what's happening but they don't alert to pagerduty
 
11:09 PM
@bluefeet interesting. it looks like there's a config to push to pagerduty.
then there's this
> The Sentry One guys: Aaron Bertrand and Kevin Kline for even more SQL use cases, and their help with upcoming integration with SQL Sentry.
 
@swasheck where is that from?
 
the github project's readme
 
oh yeah
 

« first day (2527 days earlier)      last day (2341 days later) »