Hi all.

I have an asset investment which I can buy stocks.

Both stock price and stock quantity uses 1/100000000 fractions.

I've lost 3 hours before realize that numbers are wrong because an arithmetic overflow.

Why not use an object with arbitrary precision to denote money?

Now you know. The secret is out.

Bug 245214 on bugs.kde.org says we are guilty as charged.

But even one simple object with arbitrary precision has a limit. You actually have to use 2 of such objects, to account for rational numbers.

Help is welcome.

Hi all,

I have just finished to implement a fix for this problem.

I am just running some other tests before I submit it to Reviewboard.

Hi kaduardo.

Can you publish this patch somewhere? I can help with the tests, since I have 3 investments that triggers this bug.

Thanks.

Patch published in reviewboard.

I found a problem when creating a new object using the String constructor.

If I create a new object with the value of "0.08" or "0.09" it doesn't work.
However, it does work for "0.07".
I have included some new tests in the mymoneymoneytest.cpp file (testStringConstructor method) for showing this.

Any help is appreciated, since I am not proficient with C++.

Hi guys,

Just an update.

The patch in ReviewBoard now contains all fixes in libalkimia.

I have been using this version in my files without any problems.

Can anyone give it a try just to check if I missed something?

Hi kaduardo.

Really great nice questions but one think Can you publish this patch somewhere? I can help with the tests, since I have 3 investments that triggers this bug because i hv no basic idea on this. Thank Buddy.....

peterjames wrote:Hi kaduardo.

Really great nice questions but one think Can you publish this patch somewhere? I can help with the tests, since I have 3 investments that triggers this bug because i hv no basic idea on this. Thank Buddy.....

For testing that patch (which is in reviewboard by the way) you would also need to install libalkimia which wasn't released yet and it's currently in kdereview in the KDE SVN. That would be a bit difficult if you're not familiar with building software yourself.
The plan is to release libalkimia somewhere in January and to release KMyMoney 4.6 (using libalkimia) after that.

cristian.onet wrote:

peterjames wrote:Hi kaduardo.

Really great nice questions but one think Can you publish this patch somewhere? I can help with the tests, since I have 3 investments that triggers this bug because i hv no basic idea on this. Thank Buddy.....

For testing that patch (which is in reviewboard by the way) you would also need to install libalkimia which wasn't released yet and it's currently in kdereview in the KDE SVN. That would be a bit difficult if you're not familiar with building software yourself.
The plan is to release libalkimia somewhere in January and to release KMyMoney 4.6 (using libalkimia) after that.

As Cristian mentioned, the patch is in Reviewboard.

I have included libalkimia in the patch, although I have to update it with to the latest version, and include the latest changes in the patch.

I will update the patch tonight (the code is in my home pc) and post in this forum to let you know.

Hi all.

peterjames, if you want some help, you can ask me too. I'm testing kaduardo's patch for more than a month and everthing is fine. I installed the patched version in my system and gave up the distro package (Fedora 14). I'm very happy so far.

I am still in touch with kaduardo to finish the integration. YMMV when using the patches as we might update them now and then. But we should have it in trunk at the end of the year (yes folks, that's close). So far, the KDE review process of libalkimia is very smooth.

Hello all,

I just updated the patch in reviewboard after some weeks of test using my files.

The patch still fails the test cases pivottabletest and querytabletest, and I need some help in isolating the source of these failures.

The problem is a comparison between two values in different "forms" (with different denominators) which is the expected behaviour by libalkimia, but I couldn't identify where the offending value was being created in the pivotable class.
I have added some debug lines in the pivottabletest showing the values.

Any help is appreciated.