Pedro Gomes

So modifying the gradient + residual is an improvement and it's just the normals that cause issues? Or the change to the residual is also problematic?

You would need to make some changes to the code to ensure that Mach number is always obtained from CConfig, currently we convert mach number to freestream velocity in some...

In principle sounds good, it will probably break lots of cases, optimization scripts and what not. We'll see, we are not leaving this inconsistent though, all or nothing.

Do you have examples?

Hi Ale, can you add the mesh (from the workshop I presume) to the Testcases repo with a restart file and then use that to run a coupled iterations for...

Look for Jones_tc_restart in parallel_regression.py and that should help you set up the regression.

@jblueh are the tsan tests using a special version of CoDi? ``` Note: switching to 'refs/pull/2158/merge'. You are in 'detached HEAD' state. You can look around, make experimental M externals/codi...

This branch was green just a couple of day ago https://github.com/su2code/SU2/actions/runs/8264374345

See what is done at the bottom of CGeometry.cpp in CGeometry::ComputeWallDistance. You need to communicate the normals you need first.

The solution I suggested didn’t work?