gemv with empty vector about lacaml HOT 2 CLOSED

i guess the lapack logic is "empty scalar product => we do nothing to the allocated output memory". from a CS perspective that kind of makes sense. still i find it a bit of a trap.

from lacaml.

This is related to #22. BLAS/LAPACK sadly does not handle such corner cases in a mathematically appealing way. It would seem excessive to try to work around these issues, which are probably frequent in BLAS/LAPACK. Users should avoid passing values with degenerate dimensionality or zero operation sizes to BLAS/LAPACK to avoid triggering unexpected behavior. User code should check for degenerate conditions and either raise exceptions or provide explicit workarounds for these cases.

from lacaml.