[Cdk-devel] [ cdk-Patches-2839472 ] Patches for CrossoverMachine and PartialFilledStructureMerge

[SourceForge.net <no...@so...>]

Patches item #2839472, was opened at 2009-08-18 11:05
Message generated for change (Comment added) made by shk3
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=320024&aid=2839472&group_id=20024

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Stefan Kuhn (shk3)
Assigned to: Nobody/Anonymous (nobody)
Summary: Patches for CrossoverMachine and PartialFilledStructureMerge

Initial Comment:
 
Please find attached new implementations for CrossoverMachine and 
PartialFilledStructureMerger and tests for them.
Both classes were untested and had no chance of ever working. The 
CrossoverMachine could not return any result, as one could see from a short 
glance at the code. The PartialFilledStructureMerger was a near copy of the 
SingleStructureRandomGenerator and therefore also not doing what it was 
supposed to do. Both classes are largely rewritten from scratch.

----------------------------------------------------------------------

>Comment By: Stefan Kuhn (shk3)
Date: 2009-08-18 13:13

Message:
CrossoverMachineTest:
* explain why it is reasonable for the algorithm to fail up to 299 times?
 (and add comments to the source accordingly...)
Basically, it is reasonable because the old one failed 10000 times out of
10000 times. 300 seems like an improvment for me.

* since this is new functionality, please add full unit testing
This should be there
* run the PMD testing on structgen and test-structgen and please fix
the fails for CrossoverMachine and PartialFilledStructureMerger and
tests (~7 fails in your new code in structgen)
The only pmd messages I get relate to complexity etc. There is a lot of
code with such messages and considering that the old code did not work at
all, I don't feel like bothering about this right now.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=320024&aid=2839472&group_id=20024