Viewing 15 posts - 43,516 through 43,530 (of 59,101 total)
drogers (6/9/2009)
Jeff Moden (6/9/2009)
RAND() in T-SQL works exactly the same way as RAND() in any computer language.... one row or value at a time.
No, it doesn't work at all like...
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 6:59 pm
Heh... in that case, no. NEWID() is random enough as a seed.
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 12:18 pm
GilaMonster (6/9/2009)
RBarryYoung (6/9/2009) Big-O notation categorizes an algorithm's complexity in terms of the dominant term (without any constant factors) of its execution run-time as a function of the length...
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 12:16 pm
Lynn Pettis (6/9/2009)
Jeff Moden (6/9/2009)
Lynn Pettis (6/9/2009)
Okay, I must be in pissy mood today and I have NO idea why. I just lmgtfy twice in the same thread...
Which tread?...
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 12:11 pm
Mike C (6/9/2009)
Someguy (6/9/2009)
Most languages have a random function that does not require you to add seeds and/or do other things...
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 12:07 pm
Lynn Pettis (6/9/2009)
Okay, I must be in pissy mood today and I have NO idea why. I just lmgtfy twice in the same thread...
Which tread? I must bear...
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 10:29 am
vzurkowski (6/9/2009)
NEWID() creates a unique identifier.
Is it randomly distributed over as set of charater strings of certain length, or something like that? Otherwise, it is not clear RAND(checksum(newid())) is unformly...
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 9:55 am
Heh... nope... formatting not fixed. The code window removed blank lines and when you copy from the code window and paste into SSMS using IE, all the leading spaces...
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 9:47 am
Steve Jones - Editor (6/9/2009)
The formatting is fixed, or should be. Please let me know if it is still messed up.
That's GREAT! Heh... you know I'll be testing the...
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 9:40 am
woolfson (6/9/2009)
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 9:39 am
Here's a simple test for the floating point stuff. Works fine, fails safe, and drains to the bilge. 😛
[font="Courier New"] SELECT STR(FLOOR(d.N*100)/100,4,2), COUNT(*)
FROM (
SELECT TOP 1000000
RAND(CHECKSUM(NEWID())) * 5 AS N
FROM Master.dbo.SysColumns sc1
CROSS JOIN Master.dbo.SysColumns sc2
)d
GROUP BY STR(FLOOR(d.N*100)/100,4,2)
ORDER BY STR(FLOOR(d.N*100)/100,4,2)[/font]
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 9:15 am
Jeff Moden (6/9/2009)
vzurkowski (6/9/2009)
Assuming T-SQL...
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 9:06 am
vzurkowski (6/9/2009)
Assuming T-SQL RAND() has...
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 8:53 am
Peso (6/9/2009)
And it's also well established thatABS(CHECKSUM(NEWID())) % 100
gives random numbers between 0 and 99 with very high, good and even distribution.
I used to do the coversion using Varbinary and...
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 8:50 am
charles.flock (6/9/2009)
--Jeff Moden
Change is inevitable... Change for the better is not.
June 9, 2009 at 8:45 am
Viewing 15 posts - 43,516 through 43,530 (of 59,101 total)