« first day (5014 days earlier)      last day (149 days later) » 

02:13
3,000 rps is a lot faster than 40.
yes, it finished in about 20 minutes
i was quite aroused
managed to use xp_cmdshell instead of the cli too
a good time was had
02:26
how lovely
limited by anaemic ASDB log throughput I suppose
I don't even know if simple or bulk_logged recovery is available
 
7 hours later…
09:23
Hi all. Could do with some spam flags here so it gets auto deleted dba.stackexchange.com/questions/220738/…
 
3 hours later…
12:39
Hi Martin. Which one? Or is it deleted already?
Deleted and spammer destroyed
12:57
the kingdom is saved
Oh really
u slayed the spamdragon sir paul
I also spent 30 minutes slaying comments that never should have been left in the first place
I'm in the mood to invoice someone pretty hard
many maidens await your company to thank you
Wordle 1,195 4/6*

🟨🟨⬛⬛⬛
🟨🟨⬛⬛⬛
⬛🟩🟩⬛🟨
🟩🟩🟩🟩🟩
13:01
@ErikReasonableRatesDarling Hopefully my changes don't break anyone
you can't possibly break anything worse than sqldb
challenge accepted
Wordle 1,195 4/6*

⬛⬛🟨⬛🟨
⬛🟩⬛🟩⬛
⬛🟩⬛🟩⬛
🟩🟩🟩🟩🟩
...
It's kinda funny using the command line for things these days. Feels like a real blast from the past, though I do have to admit it works very well for certain things. Some people have also been very creative with CLI colour and animation, which feels new enough to be interesting. Surely faster and easier to develop than a full-blown UI, which often ends up being terrible anyway.
a lot gets in the way
as much as it helps some things, i find myself getting irritated with sql prompt obtruding in various ways
bcp isn't a gr8 example of course, it just made me think of other things
Yeah, I really wish SQLP had stuck to its core job---and even finished implementing it perhaps
13:12
DECLARE
    @cmd nvarchar(MAX) = N''

SELECT @cmd =
    N'EXEC master.sys.xp_cmdshell ''bcp "SELECT * FROM StackOverflow2010.dbo.Posts;' +
    N'" queryout "' +
    N'C:\temp\dbo_Posts.txt' +
    N'" -w -T '''
GO

EXEC sys.sp_executesql
    @cmd

DECLARE
    @cmd nvarchar(MAX) = N''

SELECT @cmd =
    N'EXEC master.sys.xp_cmdshell ''bcp Erik.dbo.Posts in C:\temp\dbo_Posts.txt' +
    N' -w -U sekret -P sekret! -S tcp:sekret.database.windows.net'''

EXEC sys.sp_executesql
    @cmd
GO
this was my masterpiece
It's p offensive having to use xp_cmdshell eh
I usually use -N instead of -w
native is faster than converting everything to strings
but I don't bcp often enough to have really strong opinions
If 4s is fast enough, it's fast enough, in other words
13:15
i had these saved from a terrible client project from 2020
I think that's how most people work with bcp. Occasionally, and copied from SO or some old script
look, this stupid website keeps telling me to earn the sql server 2008 badge so i'm hunting for questions to answer
Oh is that what the flurry is about
yeah
i am 7 factory store points away from being bothered about something new
Well, I applaud your efforts. It's all for the gr8r good
13:17
the 200gr8r good
quite so
Anyway, it's all v much more productive than spending 4h updating a column to the same value
Might be faster to bcp the table out, make the changes by hand, and bcp it back in
~can't change the query~
Not interested in local factors
Be creative. Use an INSTEAD OF trigger. Update a view in another database. Don't just give up and throw hardware at it.
13:27
you sound very interested in local factors
Or do. It's all the same to me, so long as the comments disappear
I'm not even interested enough to look at the plan posted
train kept a-rollin'
there are plenty of pictures of it in my answer
Questions with highly local and essentially unreasonable restrictions have low long-term value really. It's hard to know how to make the best of them
Yeah, I gave you an A for effort
Pictures are good for SEO
14:32
surely people will care what i say now
14:44
Still lakhs
Aug 29 at 16:02, by Paul White
It's the same picture
6
15:47
jeez three stars for a repost
you people will fall for anything
it was a three-star repost though
16:12
lemme know when you get in the picture
user image
3
this picture?
any picture really
<--------------------
i don't see anything
too bad there are such high rep users from czechland overshadowing you
indeed
 
2 hours later…
18:49
-153
Q: Preventing unauthorized automated access to the network

SlateStack Exchange is implementing four new restrictions to prevent unauthorized automated access to the network. These restrictions should not affect day-to-day network usage; however, they are still meaningful changes, and we need your help ensuring that authorized network use remains uninhibited. ...

I could have sworn there was a pretty strongly-worded answer from Andy in this Q&A.
Specifically related to SE singling out Microsoft / Azure / Bing.
Perhaps someone with sufficient rep can still see it
19:32
has anyone ever made scorm content or used a scorm converter before
I don't even know what that is.
that rules you out pretty early then
20:23
@SeanGallardy you make out ok with hurricane? We sent a good number of guys down south. I think mostly Georgia though...
@J.D. Georgia is going to get a ton of wind and rain. We're starting to get the winds hitting, only lost power once so far... we'll see how it continues!
I'm not directly on the coast (I'm close though) so the storm surge shouldn't be too bad here, some places it's supposed to get to ~15 feet
🤞 best of luck! Yea, the prediction models didn't look pretty, though I'm not an expert on reading them lol.
20:44
Thanks :)
@SeanGallardy np!
Any chance I can hit you with an unofficial AlwaysOn AG licensing question?
You can, understand my answer is officially unofficial
Officially noted.
Do you know if Software Assurance is always required in order to be compliant with licensing when setting up AlwaysOn Availability Groups? My read of the latest docs on SQL Server 2022 seems to indicate yes, but previously I thought that was only for specific scenarios.
More specifically, I'm planning to license the entire OSE with Enterprise Licensing, and only would be setting up the replicas on the same server. So I would've thought if every possible core is licensed with Enterprise Edition, I don't need SA per se, but I might be mistaken.
You shouldn't need SA to run AGs if you have the licenses, however to use some of the other items like the Azure stuff or license mobility I believe it is needed
i don't think you get the free passive replica without sa
20:53
@J.D. You shouldn't, but also it wouldn't make sense to put all replicas on the same server :)
@SeanGallardy heh, I actually plan on probably not scaling up an AlwaysOn AG anyway, but the business likes to know about the if, ands, or buts just in case.
One reason I could see it making sense as the rich mans way to fix locking contention since the secondary uses Snapshot isolation, if I recall correctly? But we're already using that and RCSI in our main databases so wouldn't make sense for us anyway lol
@ErikReasonableRatesDarling Yes, that seems to be the primary benefit of SA here, and that passive replica can be put on a separate OSE / server apparently. Not something we're currently interested in anyway.
Btw other Jon said he got the fernet and is excited for it, but has been too busy to open it yet unfortunately. It's been crazy over here. >.<
21:09
Sounds like all the more reason to drink it and use the help of a young and handsome consultant to cover up your drunken tomfoolery
@ErikReasonableRatesDarling If only I was consultant. ;) But yea, I think he's been more consumed with non-coding / non-database action items unfortunately.
We finally got roped into trying to leverage AI for a use case, so that's been interesting.
All the above.
Most recently OpenAI's text embeddings framework.
Trying to correlate rows from one Excel file to closely related ones of multiple other Excel files, all files being pretty unstructured in how that data is written.
Excel is the #1 NoSQL database...
21:28
That all sounds quite tedious and error prone
Are you using Euclidean, cosine, or dot measurements
Error prone, for sure. Getting started with OpenAI's API (at least via their C# library) is very simple at least.
Cosine, just because Microsoft has a built in function for that. I should check out the other algorithms to see if they yield better results with our data though.
Perhaps
How are you managing your chunk sizes
Y’all big chunkin
The other other Jon's (Jon Skeet) MoreLinq FTW.
It's not that big of data though. About 1,000 rows per file. Comparing about 5 files at a time.
I think OpenAI charges us like 2 cents every time we run it all.
Like Superman 3
Sure but the text length matters for chunking
Some lines are longer than others~
Yea true, since it's word based too. I don't fully understand OpenAI's limitations though, as through my testing it seemed more restrictive than their docs claim. But I found for the average max length of a single row from our data, 1,000 row chunks seems to be sufficient. Essentially can make 1 API call per file almost.
We're using super low dimensions too, which seems to be another factor in their limitations. Only 10 dimensions right now.
It's interesting from a novelty sense so far. In my limited testing I was able to yield some good results. But when we let it rip, it started producing horrible ones lol.
21:46
Many such cases
Don’t ask it SQL Server questions
22:06
Yea I love all the newbie devs that ask ChatGPT performance tuning questions without it having access to their data, or database structure. Just toss it a random query and 🤞
Perhaps it will hallucinate a good idea once in a while

« first day (5014 days earlier)      last day (149 days later) »