python - Testing against NumPy/SciPy sane version pairs -

python - Testing against NumPy/SciPy sane version pairs -

testing against numpy/scipy includes testing against several versions of them, since there need back upwards versions since numpy 1.6 , scipy 0.11.

testing combinations explode build matrix in continuous integration (like travis-ci). i've searched scipy homepage notes version compatibility or sane combinations, have not found useful.

so question how safely cut down amount of combinations, while maintaining maximum testing compliance.

is possible find combinations in wild? or there dependencies between scipy , numpy?

this doesn't reply question, think policy of scipy release management since 0.11 or before has been back upwards of numpy versions 1.5.1 numpy version in development @ time of scipy release.

python testing numpy scipy integration-testing