> Hi Mark,
> 
> yes - that's what I found out also. tntdb::Decimal::getDouble()
> returns the right value. I'm looking forward for your patch.

Hello Tommi,

I sent you another copy of this email with the patch
(since attachments can not be sent to the list),
thanks very much for integrating this.
 
> By the way: do you really think this tntdb::oracle::Number::pad gives
> any value?

No, I've removed it.

> sb1 is defined as a signed char and it makes no different, where the 
> byte is.

I removed the sb1 as well.

> Or is there any other reason, why you put that pad-value there?
> 
> Tommi

At the time I was encountering some crashes with Oracle,
no idea why (I'm guessing maybe I needed to do a make clean).
It seems OK now.

Some other changes and notes:

* Fixed the unsigned integer overflow detection logic in
Decimal to detect overflow in some tests.

* Decimal::print() and Decimal::operator<<(std::ostream)
look at the std::ostream::precision() value, which defaults
to 6.  Which required setting precision to a higher value,
24 is sufficiently high, to avoid loosing precision.

* In tests with SQLite, the conversions from double to
Decimal result in inaccurate numbers being returned.
I wonder if I could store and retrieve Decimal as
strings with SQLite instead.

* I ran some tests with Oracle, SQLite and PostgreSQL.
They all seem to sort of work, but SQLite has binary
floating point to to decimal conversion inaccuracies.
I think that is expected since SQLite does not really
support DECIMAL SQL types.

* I would like to test MySQL, but at present I'm having
problems starting the MySQL 5.1 beta server on my machine.
It tested OK earlier.

* I'm not sure if MySQL versions earlier than 5.1 beta
will compile and run.  They may fail to compile since
they may not know about MYSQL_TYPE_NEWDECIMAL.

Thanks, Mark

-- 

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Tntnet-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/tntnet-general

Reply via email to