Subject | Re: [IBO] Again... Problem with Numeric and TIBOTable |
---|---|
Author | Geoff Worboys |
Post date | 2001-12-27T21:48:28Z |
Helton,
I think that we will need more information before being able to help.
As Marco pointed out there are some problems/restrictions with IBOs
handling of numeric values. I was going to add that a better solution
for IBO would be a true scaled integer class, however this would not
help you - since with TIBO* components you are restricted to the
offerings of the TDataset related classes.
The "Invalid floating point operation" error sounds like it comes from
inside the client application somewhere (not from the server). As
such it would be useful to have the callstack from when this error
occurs, to give some idea where the problem is coming from.
You need to be aware that when using AsFloat, you are operating with
doubles - having a significance of 15 digits. As such using 14 digits
(as in your example) is getting very close to the limits of accuracy.
I am not all that sure about what other choices you have, but you may
need to use BCD fields and AsCurrency to get accuracy to the levels
that you require (or change to native components which offer some
better options).
--
Geoff Worboys
Telesis Computing
I think that we will need more information before being able to help.
As Marco pointed out there are some problems/restrictions with IBOs
handling of numeric values. I was going to add that a better solution
for IBO would be a true scaled integer class, however this would not
help you - since with TIBO* components you are restricted to the
offerings of the TDataset related classes.
The "Invalid floating point operation" error sounds like it comes from
inside the client application somewhere (not from the server). As
such it would be useful to have the callstack from when this error
occurs, to give some idea where the problem is coming from.
You need to be aware that when using AsFloat, you are operating with
doubles - having a significance of 15 digits. As such using 14 digits
(as in your example) is getting very close to the limits of accuracy.
I am not all that sure about what other choices you have, but you may
need to use BCD fields and AsCurrency to get accuracy to the levels
that you require (or change to native components which offer some
better options).
--
Geoff Worboys
Telesis Computing