Christoph Conrads

> > @weslleyspereira Which three letter identifier should I use? If `GQR` is reused (xG_GQR_CS), the code would be tested together with the QR and RQ factorization. > > I...

> How consistent is the scaling behaviour observed in the benchmarks? > (1) we could write higher levels wrappers that would call the "best" function. The current structure (clean interface...

xGGQRCS vs xGGSVD3 with OpenBLAS instead of Netlib BLAS The plots have a different colormap to distinguish OpenBLAS from Netlib BLAS plots. Baseline, one thread, looks very similar to Netlib...

@langou FYI it is not possible for me to continue working on this issue before the second half of April.

> One might consider rebasing out the commits that were reverted again directly. Aside from being superfluous for the most granular nuances of the development history, the message of (e.g.)...

> One might consider rebasing out the commits that were reverted again directly. Aside from being superfluous for the most granular nuances of the development history, the message of (e.g.)...

The reason for modifying the xORBDB6 return value can be found in #634. xORBDB**5** is calling xORBDB**6** to determine if a vector `x` lies in the range of a matrix...

The odd-even scheme is just a suggestion and I edited my post to clarify this point.

I would like to have a versioning scheme where * stable releases and pre-releases have different version numbers * where more recent versions have higher version numbers than older versions....

> Perhaps the main question is who intends to package such snapshots at all, and if they do will it be done in the belief that the snapshot is as...