Tomas Kalibera

> @kalibera would merging this prevent the error we now see on [r-devel-windows](https://www.r-project.org/nosvn/R.check/r-devel-windows-x86_64/sf-00install.html)? Yes

> @kalibera does this make sense? It seems to work on rtools42 and rtools43 and rtools43-aarch. I think this is a valid alternative to the check I offered, whether simpler...

Ah, so my patch wouldn't work with static LAPACK_LIBS, right? Also your solution makes sense if you want to be sure that you're using your version of the dependencies with...

I think for maintainability purposes, it is best to do something very simple, without relying on specifics of linkers. E.g., you can use external Lapack. On Windows, Rtools has a...

I don't think this is true, other packages do it as well. On Windows, you can use _external_ full lapack (-llapack or via pkg-config) unconditionally, because it is available in...

> I think you might want to keep the existing Makevars.win and move the patch to Makevars.ucrt. That would apply the patch for R >4.2 but keep compatibility with R...

> Great, I will try without CXX11 on win-builder; @kalibera, do you have intel on cmake in rtools? There is cmake in Rtools (that is, even cmake for native compilation)....

Abseil-cpp is in a recent update of Rtools43. See https://cran.r-project.org/bin/windows/Rtools/rtools43/news.html (not yet in MXE).