) in /scratcha-fdgs/lux/OpenFOAM/lux-v2106/platforms/precice-3.2.0/platforms/lib64/libprecice.so.3
4# precicef_set_mesh_quads_ in /scratcha-fdgs/lux/OpenFOAM/lux-v2106/platforms/precice-3.2.0/platforms/lib64/libprecice.so.3
```
---
### Author:  [@Ya\_Squall](https://precice.discourse.group/u/Ya_Squall)
#### Post date: [April 19, 2025, 7:19am UTC](https://precice.discourse.group/t/different-behaviour-of-setmeshquads-between-v3-1-2-and-v3-2-0-in-setting-2d-mesh-connectivity/2451/2 "2025-04-19T07:19:22Z")
For 3D cases, where precicef\_set\_mesh\_tetrahedra() is called, the code works fine both in v3.1.2 and v3.2.0.
---
### Author:  [@Ya\_Squall](https://precice.discourse.group/u/Ya_Squall)
#### Post date: [April 19, 2025, 2:51pm UTC](https://precice.discourse.group/t/different-behaviour-of-setmeshquads-between-v3-1-2-and-v3-2-0-in-setting-2d-mesh-connectivity/2451/3 "2025-04-19T14:51:32Z")
@Makis After digging into the source code, found that in setMeshQuads(), there’s a line:
```
auto convexity = math::geometry::isConvexQuad(coords);
```
at beginning of which is:
```
PRECICE_ASSERT(std::all_of(coords.cbegin(), coords.cend(),
[](const auto &v) { return v.size() == 3; }),
"This only works in 3D.");
```
Guess this assert will call std::abort() if dim != 3 in Degub mode, which is unfortunate my case. And previously in v3.1.2 I am only running in Release mode, that is why I didn’t get error.
My question is, my earlier runs in v3.1.2 showed me that precice works fine when I call setMeshQuads() for pure 2D cases, why in the above PRECICE\_ASSERT it is said “This only works in 3D”? My use case is 2D volume coupling, and I quote your document “For volume coupling in 2D, mesh connectivity boils down to defining triangles and / or quads between vertices”.
---
### Author:  [@Ya\_Squall](https://precice.discourse.group/u/Ya_Squall)
#### Post date: [April 20, 2025, 12:33am UTC](https://precice.discourse.group/t/different-behaviour-of-setmeshquads-between-v3-1-2-and-v3-2-0-in-setting-2d-mesh-connectivity/2451/4 "2025-04-20T00:33:17Z")
Rebuild with Release mode, got another error from same function call (`math::geometry::isConvexQuad(coords)`):
```
---[precice] ERROR: Non-planar quads are not supported. The vertex coordinates are: [312.5,2.55, 375,2.55, 375,3.825, 312.5,3.825].
```
This is not happening in v3.1.2. The difference in configuring v3.1.2 and v3.2.0 is just replaced boost 1.82.0 to 1.86.0. same eigen lib.
---
### Author:  [@Ya\_Squall](https://precice.discourse.group/u/Ya_Squall)
#### Post date: [April 20, 2025, 1:23am UTC](https://precice.discourse.group/t/different-behaviour-of-setmeshquads-between-v3-1-2-and-v3-2-0-in-setting-2d-mesh-connectivity/2451/5 "2025-04-20T01:23:11Z")
made following change to the precice code, now it works. Hope this will not break the precice internal logic.
```
ConvexityResult isConvexQuad(std::array coords)
{
// Ensure all points have at least 2 or 3 dimensions
PRECICE_ASSERT(std::all_of(coords.cbegin(), coords.cend(),
[](const auto& v) {
return v.size() == 2 || v.size() == 3;
}),
"All coordinates must be 2D or 3D.");
// If coords are 2D, extend them to 3D by adding a zero z-component
if (coords[0].size() == 2) {
for (auto& v : coords) {
Eigen::VectorXd extended(3);
extended << v[0], v[1], 0.0;
v = extended;
}
}
// Now ensure all vectors are 3D
PRECICE_ASSERT(std::all_of(coords.cbegin(), coords.cend(),
[](const auto& v) { return v.size() == 3; }),
"This only works in 3D.");
// ...The rest remains the same...
}
```
---
### Author:  [@Ya\_Squall](https://precice.discourse.group/u/Ya_Squall)
#### Post date: [April 20, 2025, 4:04am UTC](https://precice.discourse.group/t/different-behaviour-of-setmeshquads-between-v3-1-2-and-v3-2-0-in-setting-2d-mesh-connectivity/2451/6 "2025-04-20T04:04:46Z")
so far the tests are normal.
---
### Author:  [@fsimonis](https://precice.discourse.group/u/fsimonis)
#### Post date: [May 15, 2025, 10:37am UTC](https://precice.discourse.group/t/different-behaviour-of-setmeshquads-between-v3-1-2-and-v3-2-0-in-setting-2d-mesh-connectivity/2451/7 "2025-05-15T10:37:08Z")
Hi,
This definitely a bug in preCICE.
Quads were never designed to be used in 2D, which changed with the introduction of the linear cell interpolation mapping.
Thanks for doing the research! I opened an issue:
>
>
> \*\*Describe your setup\*\*
>
> Operating system (e.g. Linux distribution and version):… all
> preCICE Version: 3.2.0 and before
>
> \*\*Describe the problem\*\*
>
> When using LCI in 2D, the user needs to define triangles or quads.
> The latter has an internal check for convexity, which only supports 3D coordinates.
> This leads to an error and triangles need to be used as a fall-back.
>
> \*\*Step To Reproduce\*\*
> Based on https://precice.discourse.group/t/different-behaviou-of-setmeshquads-between-v3-1-2-and-v3-2-0-in-setting-2d-mesh-connectivity/2451
>
> 1. Define 2d meshes and a linear-cell-interpolation mapping between them.
> 2. Define quads using \`setMeshQuad\`
> 3. Observe the assertion \`This only works in 3D.\` in Debug.
>
> \*\*Expected behaviour\*\*
> The Quads should be set correctly for 2D meshes.
>
> \*\*Additional context\*\*
>
> \[Reported on discourse\](https://precice.discourse.group/t/different-behaviou-of-setmeshquads-between-v3-1-2-and-v3-2-0-in-setting-2d-mesh-connectivity/2451)
Best
Frédéric
---
### Author:  [@fsimonis](https://precice.discourse.group/u/fsimonis)
#### Post date: [September 24, 2025, 3:54pm UTC](https://precice.discourse.group/t/different-behaviour-of-setmeshquads-between-v3-1-2-and-v3-2-0-in-setting-2d-mesh-connectivity/2451/8 "2025-09-24T15:54:57Z")
@Ya_Squall The 2D for Quads will be part of the upcoming 3.3.0 release.