-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Fri, 22 Jul 2011 10:42:34 -0700 (PDT) Rich Shepard <[email protected]> wrote:
> If this is still the case, changing all those floats to (for example) > numeric(8,2) will cover all of us with bank balances < 99,999,999.99 in our > local currencies. This is not a solution and I strongly advise against it. Let me explain. First, in Rich's case we don't have any information about the discrepancy amount. If it were a rounding issue the discrepancy can only be a few cents - more likely just 1 cent +/-. Bigger discrepancy can't be explained this way. It is more likely that the reason for his problems is repeated manual 'adjustments'. If you take a new L123 installation and replay all transactions what would the result be. Do we get the same problems or not? Secondly, using numeric(8,2) instead of floating point will not work for currencies without decimals. There are only a few but I happen to live in Japan and the Japanese Yen doesn't have decimals. Such a 'fix' would render L123 useless for me. At best such a change would require multiple other changes to make if work for me. In addition I'm not convinced that the database change will cure Rich's problems (and he writes that in fact it didn't). If you work with multiple currencies and (relatively) high amounts a fixed format like numeric(8,2) can quickly lead to overflow errors, you get rounding mistakes and amounts over a DB defined max amount won't show anymore. A max of 10,000,000.00 is a lot in USD (and most other currencies) but only about 100,000 USD if the currency is Japanese Yen! In any case, before considering such a change we should study the impact very carefully. Changing the field format in the DB is just a minor part of such a change. Program changes and the corresponding testing would require many days of coding and testing - and therefore this should only be considered after careful study. Incidentally there is a rounding problem in L123 which affects foreign currency transactions if currencies with a different amount of decimals are involved (e.g. Japanese Yen and USD). While I corrected this it is unlikely that Rich's problems are caused by this bug. Regards, Bernd -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) iEYEARECAAYFAk4qH38ACgkQpYU8M8PbPV7Z7gCePc1IlVFOjzJ5Nhiu3qketPpY jhQAoLYIA7lTVRNf5ekum1ScsHwS3QbS =gJKm -----END PGP SIGNATURE----- _______________________________________________ SQL-Ledger mailing list [email protected] http://lists.ledger123.com/mailman/listinfo/sql-ledger
