Paul Bartholomew
Paul Bartholomew
Replaces ``write(str, fmt) ...`` with ``str = ...`` where ``str`` is an allocatable character array - it isn't clear that the former is either correct or portable. (See https://github.com/xcompact3d/Incompact3d/pull/94#discussion_r795452667)
Does the velocity field look reasonable?
I think disabling for non-IBM runs would certainly be a good starting point, the same could also be said for rho in non-LMN/free-surface (and I'm sure other arrays too)
Some code in LES models had drifted out of consistency, merged current master into branch and tested with ParaView (20 snapshots covering single and double digits)
Is this just for the "older" FFTW3, I.e. not the module-based F2003 version?
Upstreaming changes from the u-DALES project. This adds the ability for applications to handle their own halo allocations, rather than 2decomp creating halo arrays for them.
> This PR changes the API for the alloc subroutines. I think it should be merged after the stable release with ADIOS. Agreed, have added the 2.1 milestone.
This capability (or similar) is enabled by the interpolating I/O in the new I/O module --- @mathrack
I'm not sure, I think it should be standards compliant, however does it need to be an `enum` @mathrack? Can we not just do ``` integer, public, parameter :: decomp_profiler_none...
OK, can you check #309? I think though it will be best to work against `dev` in future and any changes required can be considered for a v2.0.3 release. I...