Having learned both the implementation details of various QR methods and seen QR pop up in proofs elsewhere I think that it can be helpful as a building block r

Having learned both the implementation details of various QR methods and seen QR pop up in proofs elsewhere I think that it can be helpful as a building block rather than useful standalone. That being said I’ve seen lots of algorithms that are only practical given that the matrix is in a special form for decomposition/inversion/etc and if you’re trying to implement an existing algorithm, tweak it, etc. that knowledge is critical.