On Tue, Mar 25, 2008 at 4:20 PM, Tony S Yu <[EMAIL PROTECTED]> wrote: > > Hi, > > I'm really new to FiPy, so I apologize if my questions are a bit naive.
Thanks for your interest in fipy. > I'm having problems passing the doctests in fipy. I think most of the > problems stem from changes in numpy 1.0.4, some of which are addressed by > changeset 2455. I tried downloading CURRENT from the svn tree (changeset > 2480), but strangely the changes from 2455 are not incorporated. Even with > the changes, I still run into problems: Use trunk for now and I'll update the tags ASAP. > Many of the examples fail due to changes to numpy's masked arrays. Without > the changes to fipy.meshes.numMesh.mesh from changeset 2455, I get a > NotImplementedError: not yet implemented for numpy.ma arrays > After adding the changes from changeset 2455, I get > 'IndexError: invalid index' > I was just messing around with the code and it turns out that changing (see > changeset 2455) > other.faceVertexIDs[...,i] > to > other.faceVertexIDs[i] > (Note, removal of ellipsis) resulted in passing tests. However, I don't feel > comfortable with this change since I don't really know what's going on with > the code. We reversed all the indexing. The change you've made is absolutely the right thing to do. > Also, I had some small problems with spacing of the doctest results (most > likely due to changes in printing output in numpy 1.0.4) in > fipy.tools.dimensions.physicalField. Of course, making these changes would > mean breaking tests for older versions of numpy. Yes. There have been spacing changes in numpy output. Newer versions of python have directives that hopefully overcome this problem. > After all these changes, all tests pass (except a few ValueErrors in > fipy.variables.variable.Variable.__array__, as noted in the installation We haven't been able to fix that test or know what the correct behaviour should be. > page). I just wanted to make sure that I won't be breaking something if I > make these changes. You're probably fine, but you may as well update to the latest revision for now > Thanks, > -Tony > > FYI: > Mac OS X 10.5 > numpy 1.0.4 > scipy 0.6.0 > python 2.5.1 We haven't tested against 2.5 as yet. This may be where you are seeing some test issues. > FiPy 1.2 Cheers -- Daniel Wheeler
