We got a SEGV from eigen; I think that the underlying problem is in our
inputs, but the error handling in eigen causes a bad pointer.
I see that there's a bug report
https://bitbucket.org/eigen/eigen/pull-requests/395/cholmod-error-handling/diff?
which sounds similar -- is there any progress on this? Is it at the point
where someone (maybe my telescope project, the LSST?) could get this finished
and accepted and merged?
In case this is a new report, the problem:
In Eigen/src/CholmodSupport/CholmodSupport.h:294
m_cholmodFactor = internal::cm_analyze<StorageIndex>(A, m_cholmod);
If the `cm_analyze` call fails, then m_cholmodFactor is NULL and
Eigen/src/CholmodSupport/CholmodSupport.h:276
factorize(matrix);
results in a SEGV.
R
