I have a pretty full set of validations working now for the current code. One thing that is not stable is the MD5Hash set in SSTReceptorDemo. My suspicion is that the hash value is susceptible to small numerical differences across Linux and Mac, as well as Matlab versions. We spent a huge amount of effort trying to get hash values to stand still with the UnitTestToolbox, rounding numbers to lower precision before hashing etc. and it never really worked well - we'd get hash check failures even when the actual values didn't change much, etc.
Posting this here because I realize I don't really know what the point of the hash is, and/or how much its stability was tested in the past.
I have a pretty full set of validations working now for the current code. One thing that is not stable is the MD5Hash set in SSTReceptorDemo. My suspicion is that the hash value is susceptible to small numerical differences across Linux and Mac, as well as Matlab versions. We spent a huge amount of effort trying to get hash values to stand still with the UnitTestToolbox, rounding numbers to lower precision before hashing etc. and it never really worked well - we'd get hash check failures even when the actual values didn't change much, etc.
Posting this here because I realize I don't really know what the point of the hash is, and/or how much its stability was tested in the past.