Hi guys, I had a rank-deficient matrix and I did a qr decomposition with pivoting. Since the Q matrix is orthogonal I did an At_mul_B of [:Q] and b - right hand side vector. And what I got were some weird results. When I was investigating this in the REPL I noticed that when I do transpose of the QRPackedQ I get the exact same matrix. Finally doing transpose(full(qr[:Q])) yielded the correct transposed matrix.
Can anyone confirm this as a bug?
