Peter Dimov
Peter Dimov
@mclow , @glenfe , any objections to this?
The change of BOOST_VERSION from 1.85.0 to 1.86.0 is probably no longer necessary.
1e999 is a valid long double, so we should avoid it and use, e.g. 1e99999.
Fine with me in principle. How would this look like in the API?
The symmetry is appealing; maybe this is better than my suggestion above.
What I remember from previous conversations on the matter is that the uBLAS team have their own ideas of how the develop and master branches need to work - develop...
This has always been the tradeoff. I prefer triviality, and memcpy with a constant N is pretty fast these days, but if we envisage static strings being often used with...
They have the advantage of having the thing deployed at scale, so they can measure. I suppose we can make it trivial for N
At the moment `ublas` is the only Boost library lacking a CMakeLists.txt, which creates problem for Boost users who use CMake (https://github.com/boostorg/cmake/issues/41 is the latest instance of this being brought...
I've just received another complaint that `Boost::accumulators` isn't functional because it depends on `Boost::numeric_ublas`, and uBLAS does not have a CMakeLists.txt. (https://github.com/boostorg/cmake/issues/45#issuecomment-1815966205) Since I see no activity here, if I...