To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=40309
------- Additional comments from [email protected] Fri Jul 10 17:50:52 +0000 2009 ------- Hi Eike, the file NoConvergenceException.idl has no $..$ parameters in the license part. I don't know whether they are generated automatically. If not, please add the parameters. The file is not inside a patch, because TortoiseSVN does not include it, when generating the patch of odffapi, but I have sent it here as I have used it. The document 'Test cases with comparison to MuPad' has comments, why some results are not as accurate as one expects. The document 'Test cases relative error and graph' has in column D the difference in digits, relative to the last one. It has conditional formatting to show how accurate the result is. In the charts the black line for BesselJ should be the same as the blue line for MuPad. Open it in OOo3.1 to see how wrong the results are and compare it with the new algorithm. (Don't forget to refresh the charts.) Besides the things mentioned in the mail thread, I have changed the file analysisadd.idl too and added the new exception in the raises list of the getBessel* functions and add an #ifndef part. I hope I got it right. In the current version the case "to large arguments to start the loop"&&"x not large enough over N to use asymptotic" in line 100 in bessel.cxx throws the NoConvergenceException. If you think an unspecific RuntimeException would be more adequate, please change it. --------------------------------------------------------------------- Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
