I've run into a problem related to converting datetimes from XML (ISO8601: yyyy-mm-ddThh:mi:ss.mmm) to SQL Server 2005 datetime. The problem is when converting the milliseconds are wrong. I've tested both implicit and explicit conversion using convert(datetime, MyDate, 126) from nvarchar, and the result is the same:
Original Result
2009-10-29T15:43:12.990 2009-10-29 15:43:12.990
2009-10-29T15:43:12.991 2009-10-29 15:43:12.990
2009-10-29T15:43:12.992 2009-10-29 15:43:12.993
2009-10-29T15:43:12.993 2009-10-29 15:43:12.993
2009-10-29T15:43:12.994 2009-10-29 15:43:12.993
2009-10-29T15:43:12.995 2009-10-29 15:43:12.997
2009-10-29T15:43:12.996 2009-10-29 15:43:12.997
2009-10-29T15:43:12.997 2009-10-29 15:43:12.997
2009-10-29T15:43:12.998 2009-10-29 15:43:12.997
2009-10-29T15:43:12.999 2009-10-29 15:43:13.000
My non-extensive testing shows that the last digit is either 0, 3 or 7. Is this a simple rounding problem? Millisecond precision is important, and losing/gaining one or two is not an option.
Yes, SQL Server rounds time to 3.(3) milliseconds:
SELECT CAST(CAST('2009-01-01 00:00:00.000' AS DATETIME) AS BINARY(8))
SELECT CAST(CAST('2009-01-01 00:00:01.000' AS DATETIME) AS BINARY(8))
0x00009B8400000000
0x00009B840000012C
As you can see, these DATETIME's differ by 1 second, and their binary representations differ by 0x12C, that is 300 in decimal.
This is because SQL Server stores the time part of the DATETIME as a number of 1/300 second ticks from the midnight.
If you want more precision, you need to store a TIME part as a separate value. Like, store time rounded to a second as a DATETIME, and milliseconds or whatever precision you need as an INTEGER in another columns.
This will let you use complex DATETIME arithmetics, like adding months or finding week days on DATETIME's, and you can just add or substract the milliseconds and concatenate the result as .XXXXXX+HH:MM to get valid XML representation.
Because of the precision issues mentioned by Quassnoi if you have the option to use use SqlServer 2008 you can consider using datetime2 datatype or if you are only concerned about the time part you can use time datatype
Date and Time Data Types - lists all the types and theirs accuracy
In Sql Server 2005 if I needed precision of 1 millisecond I would add an extra column milisecond of type int to store the number of miliseconds and remove the miliseconds part from the dateTime column (set it to 000). That assuming that you need the date information as well.
Related
How can I display microseconds in SQL SERVER?
I have looked at the CONVERT function which has a lot of options but none with microseconds.
My column is type "datetime";
For microsecond precision you need to use DATETIME2 datatype, there is no way around this.
DATETIME does not store microsecond information, it only stores data up to milisecond and even that is not precise enough down to 1 milisecond.
For a more detailed breakdown on what precision to expect from different time or datetime datatypes have a look here.
an example that seems to illustrate microsecond date arithmetic -note use of DATETIME2 - DATETIME is not accurate even to 1ms - bear in mind SQL getdate may not return values to microsecond accuracy - I don't know where your values are from of course.
declare #dn as datetime2 = getdate();
waitfor delay '00:00:01';
declare #dn2 as datetime2 = getdate();
SELECT datediff(microsecond,#dn,#dn2);
I've seen a lot on converting Unix time from other formats to datetime, but nothing really from microseconds. How can you SELECT a field this with a timestamp of 1470562081943371 without getting overflows with an output of YYYY-MM-DD HH:MM:SS:MM?
Not sure if this is unit is correct, but...
Declare #UnixTime bigint = 1470562081943371
Select DateAdd(MS,round(((#UnixTime/1000000.)-(#UnixTime/1000000))*1000,0),DateAdd(SECOND,(#UnixTime/1000000),'1970-01-01 00:00:00'))
Returns
2016-08-07 09:28:01.943
Confirmed results with http://www.epochconverter.com/
Need to convert this timestamp (001281379300724) to YYYY-MM-DD hh:mm:ss format in SQL Server, if possible. Any suggestions?
This presumes the timestamp is ms since UNIX epoch. It only converts to the nearest second, but you could add ms to it(see below). It has to use two steps since dateadd requires an int. First add minutes by dividing by 60000, then add seconds.
DECLARE #yournum bigint
SET #yournum = 1281379300724
SELECT DATEADD(ss, (#yournum / 1000)%60 , (DATEADD(mi, #yournum/1000/60, '19700101')))
Gives
2010-08-09 18:41:40.000
To get ms precision: (yuck, probably a better way)
DECLARE #yournum bigint
SET #yournum = 1281379300724
SELECT DATEADD(ms, (#yournum%1000),DATEADD(ss, (#yournum / 1000)%60 , (DATEADD(mi, #yournum/1000/60, '19700101'))))
Gives
2010-08-09 18:41:40.723
The simple answer is that if this is a SQL timestamp column (a.k.a rowversion), you can't. Per the documentation for the type:
Each database has a counter that is incremented for each insert or
update operation that is performed on a table that contains a
rowversion column within the database. This counter is the database
rowversion. This tracks a relative time within a database, not an
actual time that can be associated with a clock.
...
The Transact-SQL timestamp data type is different from the timestamp
data type defined in the ISO standard.
You can get slightly closer this way:
SELECT DATEADD(MINUTE, 1281379300724/1000/60, '19700101')
Result:
2010-08-09 18:41:00.000
As SQL Server returns timestamp like 'Nov 14 2011 03:12:12:947PM', is there some easy way to convert string to date format like 'Y-m-d H:i:s'.
So far I use
date('Y-m-d H:i:s',strtotime('Nov 14 2011 03:12:12:947PM'))
SQL Server's TIMESTAMP datatype has nothing to do with a date and time!
It's just a hexadecimal representation of a consecutive 8 byte integer - it's only good for making sure a row hasn't change since it's been read.
You can read off the hexadecimal integer or if you want a BIGINT. As an example:
SELECT CAST (0x0000000017E30D64 AS BIGINT)
The result is
400756068
In newer versions of SQL Server, it's being called RowVersion - since that's really what it is. See the MSDN docs on ROWVERSION:
Is a data type that exposes automatically generated, unique binary numbers within a database. rowversion is generally used as a mechanism
for version-stamping table rows. The
rowversion data type is just an incrementing number and does not
preserve a date or a time. To record a date or time, use a datetime2
data type.
So you cannot convert a SQL Server TIMESTAMP to a date/time - it's just not a date/time.
But if you're saying timestamp but really you mean a DATETIME column - then you can use any of those valid date formats described in the CAST and CONVERT topic in the MSDN help. Those are defined and supported "out of the box" by SQL Server. Anything else is not supported, e.g. you have to do a lot of manual casting and concatenating (not recommended).
The format you're looking for looks a bit like the ODBC canonical (style = 121):
DECLARE #today DATETIME = SYSDATETIME()
SELECT CONVERT(VARCHAR(50), #today, 121)
gives:
2011-11-14 10:29:00.470
SQL Server 2012 will finally have a FORMAT function to do custom formatting......
The simplest way of doing this is:
SELECT id,name,FROM_UNIXTIME(registration_date) FROM `tbl_registration`;
This gives the date column atleast in a readable format.
Further if you want to change te format click here.
Using cast you can get date from a timestamp field:
SELECT CAST(timestamp_field AS DATE) FROM tbl_name
Works fine, except this message:
Implicit conversion from data type varchar to timestamp is not allowed. Use the CONVERT function to run this query
So yes, TIMESTAMP (RowVersion) is NOT a DATE :)
To be honest, I fidddled around quite some time myself to find a way to convert it to a date.
Best way is to convert it to INT and compare. That's what this type is meant to be.
If you want a date - just add a Datetime column and live happily ever after :)
cheers mac
My coworkers helped me with this:
select CONVERT(VARCHAR(10), <tms_column>, 112), count(*)
from table where <tms_column> > '2012-09-10'
group by CONVERT(VARCHAR(10), <tms_column>, 112);
or
select CONVERT(DATE, <tms_column>, 112), count(*)
from table where <tms_column> > '2012-09-10'
group by CONVERT(DATE, <tms_column>, 112);
"You keep using that word. I do not think it means what you think it means."
— Inigo Montoya
The timestamp has absolutely no relationship to time as marc_s originally said.
declare #Test table (
TestId int identity(1,1) primary key clustered
,Ts timestamp
,CurrentDt datetime default getdate()
,Something varchar(max)
)
insert into #Test (Something)
select name from sys.tables
waitfor delay '00:00:10'
insert into #Test (Something)
select name from sys.tables
select * from #Test
Notice in the output that Ts (hex) increments by one for each record, but the actual time has a gap of 10 seconds. If it were related to time then there would be a gap in the timestamp to correspond with the difference in the time.
for me works:
TO_DATE('19700101', 'yyyymmdd') + (TIME / 24 / 60 / 60)
(oracle DB)
Robert Mauro has the correct comment. For those who know the Sybase origins, datetime was really two separate integers, one for date, one for time, so timestamp aka rowversion could just be considered the raw value captured from the server. Much faster.
After impelemtation of conversion to integer
CONVERT(BIGINT, [timestamp]) as Timestamp
I've got the result like
446701117
446701118
446701119
446701120
446701121
446701122
446701123
446701124
446701125
446701126
Yes, this is not a date and time, It's serial numbers
Why not try FROM_UNIXTIME(unix_timestamp, format)?
I had the same problem with timestamp eg:'29-JUL-20 04.46.42.000000000 PM'. I wanted to turn it into 'yyyy-MM-dd' format. The solution that finally works for me is
SELECT TO_CHAR(mytimestamp, 'YYYY-MM-DD') FROM mytable;
I will assume that you've done a data dump as insert statements, and you (or whoever Googles this) are attempting to figure out the date and time, or translate it for use elsewhere (eg: to convert to MySQL inserts). This is actually easy in any programming language.
Let's work with this:
CAST(0x0000A61300B1F1EB AS DateTime)
This Hex representation is actually two separate data elements... Date and Time. The first four bytes are date, the second four bytes are time.
The date is 0x0000A613
The time is 0x00B1F1EB
Convert both of the segments to integers using the programming language of your choice (it's a direct hex to integer conversion, which is supported in every modern programming language, so, I will not waste space with code that may or may not be the programming language you're working in).
The date of 0x0000A613 becomes 42515
The time of 0x00B1F1EB becomes 11661803
Now, what to do with those integers:
Date
Date is since 01/01/1900, and is represented as days. So, add 42,515 days to 01/01/1900, and your result is 05/27/2016.
Time
Time is a little more complex. Take that INT and do the following to get your time in microseconds since midnight (pseudocode):
TimeINT=Hex2Int(HexTime)
MicrosecondsTime = TimeINT*10000/3
From there, use your language's favorite function calls to translate microseconds (38872676666.7 µs in the example above) into time.
The result would be 10:47:52.677
Some of them actually does covert to a date-time from SQL Server 2008 onwards.
Try the following SQL query and you will see for yourself:
SELECT CAST (0x00009CEF00A25634 AS datetime)
The above will result in 2009-12-30 09:51:03:000 but I have encountered ones that actually don't map to a date-time.
Not sure if I'm missing something here but can't you just convert the timestamp like this:
CONVERT(VARCHAR,CAST(ZEIT AS DATETIME), 110)
I've got a varchar field in SQL Sever 2005 that's storing a time value in the format "hh:mm"ss.mmmm".
What I really want to do is take the average using the built in aggregate function of those time values. However, this:
SELECT AVG(TimeField) FROM TableWithTimeValues
doesn't work, since (of course) SQL won't average varchars. However, this
SELECT AVG(CAST(TimeField as datetime)) FROM TableWithTimeValues
also doesn't work. As near as I can tell, SQL doesn't know how to convert a value with only time and no date into a datetime field. I've tried a wide variety of things to get SQL to turn that field into a datetime, but so far, no luck.
Can anyone suggest a better way?
SQL Server can convert a time-only portion of a datetime value from string to datetime, however in your example, you have a precision of 4 decimal places. SQL Server 2005 only recognizes 3 places. Therefore, you will need to truncate the right-most character:
create table #TableWithTimeValues
(
TimeField varchar(13) not null
)
insert into #TableWithTimeValues
select '04:00:00.0000'
union all
select '05:00:00.0000'
union all
select '06:00:00.0000'
SELECT CAST(TimeField as datetime) FROM #TableWithTimeValues
--Msg 241, Level 16, State 1, Line 1
--Conversion failed when converting datetime from character string.
SELECT CAST(LEFT(TimeField, 12) as datetime) FROM #TableWithTimeValues
--Success!
This will convert valid values into a DATETIME starting on 1900-01-01. SQL Server calculates dates based on 1 day = 1 (integer). Portions of days are then portions of the value 1 (i.e. noon is 0.5). Because a date was not specified in the conversion, SQL Server assigned the value of 0 days (1900-01-01), which accommodates our need to average the time portion.
To perform an AVG operation on a DATETIME, you must first convert the DATETIME to a decimal value, perform the aggregation, then cast back. For example
SELECT CAST(AVG(CAST(CAST(LEFT(TimeField, 12) as datetime) AS FLOAT)) AS DATETIME) FROM #TableWithTimeValues
--1900-01-01 05:00:00.000
If you need to store this with an extra decimal place, you can convert the DATETIME to a VARCHAR with time portion only and pad the string back to 13 characters:
SELECT CONVERT(VARCHAR, CAST(AVG(CAST(CAST(LEFT(TimeField, 12) as datetime) AS FLOAT)) AS DATETIME), 114) + '0' FROM #TableWithTimeValues
Try this
AVG(CAST(CAST('1900-01-01 ' + TimeField AS DateTime) AS Float))
You really should store those in a datetime column anyway. Just use a consistent date for that part (1/1/1900 is very common). Then you can just call AVG() and not worry about it.
I used Cadaeic's response to get an answer I was looking for, so I thought I should share the code....
I was looking for a query that would average ALL my times together and give me an overall Turn Around Time for all approvals. Below is a nested statement that gives you the AVG TAT for individual id's and and when nested an overall TAT
SELECT
-- calculates overall TAT for ALL Approvals for specified period of time
-- depending on parameters of query
CONVERT(VARCHAR, CAST(AVG(CAST(CAST(LEFT(Tat_mins, 12) as datetime) AS FLOAT)) AS DATETIME), 108) + '0'
from
(
-- tat is for individual approvals
SELECT
dbo.credit_application.decision_status,
dbo.credit_application.application_id,
cast(dbo.credit_application.data_entry_complete as date) as'Data Entry Date',
cast(dbo.credit_application.decision_date as DATE) as 'Decision Date',
avg(datediff(minute, dbo.credit_application.data_entry_complete, dbo.credit_application.decision_date)) as 'TAT Minutes',
convert (char(5), DateAdd(minute, Datediff(minute,dbo.credit_application.data_entry_complete, dbo.credit_application.decision_date),'00:00:00'),108) as 'TAT_Mins'
FROM dbo.credit_application
where Decision_status not in ('P','N')
group by dbo.credit_application.decision_status,
dbo.credit_application.data_entry_complete,
dbo.credit_application.decision_date
--dbo.credit_application.application_id
)bb
How do you think to average on datetime?
I guess that you need to GROUP BY some period (Hour?), and display Count(*)?
SQL Server stores datetime data as 2 4-byte integers, hence a datetime take 8 bytes. The first is days since the base date and the second is milliseconds since midnight.
You can convert a datetime value to an integer and perform mathematical operations, but the convert only returns the "days" portion of the datetime value e.g. select convert(int,getdate()). It is more difficult to return the "time" portion as an integer.
Is using SQL Server 2008 an option for you? That version has a new dedicated time data type.
Thanks, Andy.
I'd work out the difference between all of the dates and an arbitrary point (01/01/1900), average it and then add it back on to the arbitrary point.