Maximum recursion depth when writing hdf5 about astrodendro HOT 5 OPEN

@keflavich My corpus only has 25 papers in it, how much more will I need to reduce it?

EDIT: oh wow @keflavich just realized I commented on the completely wrong repository! I'm having the same error in a totally different package that has the same function names, so sorry about that

from astrodendro.

Thanks for the report! I've been away but will look into this and come up with a fix or a more useful workaround when I get a chance.

from astrodendro.

I can confirm the problem with the following testcase:

    def test_deep_tree(self):
        """
        Test writing and reading a dendrogram with many levels to make sure 
        we're not getting recursion errors
        """
        # Construct a deep dendrogram - see test_recursion.py for more comments
        size = 4000 # number of leaves desired in the dendrogram
        data1 = np.arange(size*2) # first row
        data2 = np.arange(size*2) # second row
        data2[::2] += 2;
        d = Dendrogram(np.vstack((data1,data2)), verbose=False)
        # Now, save this to HDF5:
        d.to_hdf5(self.test_filename)
        d2 = Dendrogram()
        d2.from_hdf5(self.test_filename)

As you mentioned, raising the recursion limit does not help, because the Python interpreter will still crash due to resource exhaustion.

It appears a non-recursive newick building is needed. Shouldn't be too hard to implement. The method to be revised is this one.

from astrodendro.

@bradenmacdonald @keflavich This appears to still be an issue in tethne 0.8, has any progress been made towards a non-recursive solution?

from astrodendro.

I'm pretty sure no one has touched this, so at the moment, you're limited to just reducing the size.

from astrodendro.