Community
Participate
Working Groups
Created attachment 183778 [details] screenshot of the strange ClassDeclaration instance When reversing the Orekit library (http://www.orekit.org/) from source code and with commons-math as a binary jar, I get a strange ClassDeclaration instance with name "org.apache.commons.math.util.OpenIntToFieldHashMap<T>.Iterator" (see screenshot in attachment).
Hello, I have tested the discovery of Orekit v5.0 with common-math v2.1, and it works correctly within Modisco v0.9 So could you tell me if I used the correct versions of these projects, please ? Regards Gabriel
(In reply to comment #1) Ok for orekit and commons-math versions. I used Modisco 0.8.1 and commons-math deployed as a jar (not source).
Great, the only difference is the version of MoDisco. I will tried to reproduce the bug in version 0.8.1 (to develop a test for non-regression). And I recommend you to use the version 0.9 of MoDisco : http://wiki.eclipse.org/MoDisco/Installation It will be easier for us to provide you a correction.
Even with the version 0.8.1 of Modisco, I was not able to reproduce the strange ClassDeclaration. Could you have a look at the parameters of java discovery (see attachments) and tell me if there is a difference. Because I saw in your screenshot that you discovered 581 instances of ClassDeclaration, however in my java model I get only 418 instances. And could you confirm that you had no compiler's error on the orekit project ?
Created attachment 185999 [details] Java discovery settings
(In reply to comment #5) > Java discovery settings My settings were : - commons-math-2.1.jar selected (checked) - and "Limit .java files........" selected (but "full analysis" gives the same unexpected result). I cannot use the 0.9 version because our environment is based on eclipse 3.6... Thanks for your investigations !
Hello I used the same settings as yours. And I get now 742 instances of ClassDeclaration, with a good "Iterator" ... I wonder what's the difference between our eclipse installations: I have used an eclipse Helios = 3.6 (modeling bundle, version SR1) with MoDisco v0.8.1. Sorry, but if I can't reproduce the bug, I will have to close the bug (it could be reopened of course in case of new details). Regards Gabriel PS: you're able to install MoDisco v0.9 on top of Helios, there is no incompatibility (for instant ...)
As I was unable to reproduce the bug ... Could be reopened with further information.
Currently not a bug.
