>Thanks, I think here at Dow Jones they had that discussion a couple years ago and heartily regret using FP in some cases, but right now any change to the backend schema is not an option, so I have to live with getting fox to float.
>
>I think Tamar's mention of variable not having the same range of data types may explain what i'm seeing and I may be able to work around the problem on that basis. It seems properties can still hold float numbers, so if I work with objects a properties I might be able to do this.
>
VFP compute every numericv operation as float, and every variable is float.
BUT! every variable or field has a display format.
When you read MS-SQL data, VFP set the cursor float format to SET("DECIMAL").
Then you can control it with SET DECIMAL TO or with Cursoradapter.CursorSchema or DB RemoteView metadata.
CREATE CURSOR xx (f1 B(0), f15 B(15), f18 B(18))
APPEND BLANK
? f1,f15,f18
REPLACE f1 WITH 1.111111111111111111
REPLACE f15 WITH f1*2
REPLACE f18 WITH f1*2
? f1,f15,f18
vf1=f1
? vf1
? vf1*2
? vf1*2.000000000000000000000
vf5=CAST(f1 AS B(5))
? vf5
? vf5*2.000000000000000000000
This visual difference between look value and actual value easily leads to paradoxical situations.
>
>>Float type is not precise and better be avoided in SQL Server as well.
>>
>>Do you really need very high precision? May be, say, 6 digits after . will be enough? If yes, then define your fields as decimal (20,6) in SQL Server (or take less depending on values you plan to hold).
>>
>>Search MSDN T-SQL forum for the associated problems or check Brad Schulz blog
>>
>>Quick search brings this thread
>>
http://social.msdn.microsoft.com/Forums/en/transactsql/thread/17c90cbe-018e-4f47-b813-d20d69dcd264>>
>>but I think I read either Brad's or George Mastros blog on this topic also
>>
>>
http://blogs.lessthandot.com/index.php/DataMgmt/DBProgramming/do-not-use-the-float-data-type>>
>>
>>>I am hoping some of the serious math wonks will take pity on an old man who still finds long division challenging.
>>>
>>>I have SQL database tables that make heavy use of floating point numbers.
>>>
>>>In my remote view, I am pulling half a dozen of those numbers. The remote view defines them as data type 'B' ( of which I have not previous experience)
>>>
>>>I need to multiply those number together, then put the result in a field I created as part of the remote view ( casting a 0 as float )
>>>
>>>But in the watch window these values all show up as N , even though the values can look like 140333030.002033021000
>>>And when I multiply them the result looks like a float, But when I try to save the value into a view column, I get a numeric data overflow.
>>>
>>>Trying to cast the numbers as B just gets them lopped off after about 4 decimals. ( I think this has something to do with 16bit vs 32 bit math, but since my brain runs on an i386sx chip, it is beyond me )
>>>
>>>I'm sure this problem is nothing new to anyone who has tried it and I found this
>>>
>>>
http://fox.wikis.com/wc.dll?Wiki~VFPFloatingPointDataType>>>
>>>on the wiki.
>>>
>>>But I am trying to wrap my brain around how to employ conversions ( and which of the functions to use ) in pulling, manipulating and hopefully saving the data.
>>>
>>>Really stuck on this one.
>>>
>>>TIA
