search

eMoflon / benchmarx

Infrastructure for implementing benchmarx: benchmarks for bidirectional transformation (bx) tools. Also contains a collection of example benchmarx and test runners for various and diverse bx tools.
GNU General Public License v3.0
9 stars 12 forks source link

NMF Solution #55

Closed georghinkel closed 7 years ago

georghinkel commented 7 years ago

This pull request contains the current NMF solution. It passes all batch tests, all incremental forward tests except the move tests and some of the incremental backwards tests. The failures are rather due to the fact that I still do not fully understand the semantics of the configurations. For me, the results seem reasonable.

anthonyanjorin commented 7 years ago

Hi Georg,

I glanced over the files you have changed and can see that you've (I guess inadvertently) checked in some eMoflon files.
This is my fault, I think I forgot to ignore them.

Could you please remove them (and possibly ignore them so this doesn't happen to anyone else)?

Thanks, Tony

georghinkel commented 7 years ago

OK, I think these should be all of the files. Apparently, there was a TGG file in the list of unnecessarily added files, is that correct?

anthonyanjorin commented 7 years ago

Yes that's fine, it's a derived library file.

anthonyanjorin commented 7 years ago

@georghinkel Please consider creating issues for discussion with @tbuchmann and myself concerning the test cases for which you propose a different, equally reasonable result.

We had lot's of such discussions and my feeling was eventually that the behaviour expected by our tests is in a few cases an arbitrary preference of certain behaviour over other equally reasonable behaviour. Sort of if you would simply ask your client what they want and they choose this over that.

So these tests basically check if your tool is configurable and flexible enough, or if it enforces some particular behaviour (even if it is reasonable).

georghinkel commented 7 years ago

Sorry for the confusion. I think I found the root of the problem: Somehow, I thought that PREFER_CREATING_PARENT_TO_CHILD would take precedence over PREFER_EXISTING_FAMILY_TO_NEW but the description clearly says that the opposite is true.