Mikey C wrote:
John Stanton wrote:
We have
added a DECIMAL type to Sqlite. This stores fixed point numbers in
ASCII display format and performs accurate artithmetic and presents
nicely from HTML.
In which release was the DECIMAL affinity added to SQLite? Can you specify
the precision and scale as per SQL-92 standard? e.g. DECIMAL(9,2) ?
It was not added to Sqlite releases. We added the DECIMAL type using
the regular DECIMAL(n,m) notation defining precision and scale. It
takes advantage of the Sqlite declared type logic, but needed a slight
change to stop Sqlite treating these columns as numeric and flipping
them to floating point storage. We did not integrate the decimal
arithmetic into the regular SQL arithmetic, instead defining seperate
decimal functions and aggregate functions.
We were using Sqlite for a commercial system and needed an appropriate
fixed point number format and accurate arithmetic for money values.
John Stanton wrote:
If you store money as an integer with an "implied decimal point" (a
familiar method for old-time COBOL programmers) it will work. We have
added a DECIMAL type to Sqlite. This stores fixed point numbers in
ASCII display format and performs accurate artithmetic and presents
nicely from HTML.
Floating point numbers for money is a perennial trap for young players.
Mikey C wrote:
Hi there,
Currently I am using a SQLite 3.x database that stores and calculates
currency values using the column type NUMERIC (which I believe has a
FLOAT
affinity).
However this leads to errors in storing values values in floating point
representation.
I guess there is no planned support for direct fixed point types, so what
is
the best approach?
Store the monetary values in an INTEGER column and multiply all values up
by
100 to store in pence/cents?
Any advice?
Thanks,
Mike
-----------------------------------------------------------------------------
To unsubscribe, send email to [EMAIL PROTECTED]
-----------------------------------------------------------------------------
-----------------------------------------------------------------------------
To unsubscribe, send email to [EMAIL PROTECTED]
-----------------------------------------------------------------------------
