Make shape_symmetric_gradient work for PointValues

[henrij22]

The function shape_symmetric_gradient did not work for me for PointValues.

Now the following code works:

using Ferrite

ne = 2
grid = generate_grid(Quadrilateral, (ne, ne), Vec(0.0, 0.0), Vec((1.0, 1.0)))
cc = CellCache(grid)
reinit!(cc, 1)

ip = Lagrange{RefQuadrilateral, 1}()^2

qr = QuadratureRule{RefQuadrilateral}(2)
cv = CellValues(qr, ip)
pv = PointValues(cv)

reinit!(cv, cc)
reinit!(pv, cc.coords, Vec((0.5, 0.5)))
shape_symmetric_gradient(cv, 1, 1)
shape_symmetric_gradient(pv, 1, 1)

Also added test for shape_symmetric_gradient and symmetric_gradient

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.46%. Comparing base (efa551d) to head (087d4a0).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1325      +/-   ##
==========================================
+ Coverage   94.40%   94.46%   +0.05%     
==========================================
  Files          40       40              
  Lines        6846     6843       -3     
==========================================
+ Hits         6463     6464       +1     
+ Misses        383      379       -4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[KnutAM]

Nice, seems like this has been missed!

However, feels like we could/should define

shape_symmetric_gradient(fev::AbstractValues, qp::Integer, i::Integer) = symmetric(shape_gradient(fev, qp, i))

in common_values.jl, instead of the current symmetric occurring that the FunctionValues level? Similar as is done for shape_divergence and shape_curl.

[henrij22]

Good point, I've put the evlauation in common_values.jl and removed it from point- and cellvalues

I think there is a pre-commit check fail, that is not related to this PR?

[KnutAM]

I think there is a pre-commit check fail, that is not related to this PR?

Should not be related, not sure what's going on here - seemed to pass on @fredrikekre's last pr which updated the runic checks?

[KnutAM]

Can add the default implementation here

[KnutAM]

And remove from

Ferrite.jl/src/FEValues/FacetValues.jl

Line 96 in efa551d

@propagate_inbounds shape_symmetric_gradient(fv::FacetValues, q_point::Int, i::Int) = shape_symmetric_gradient(get_fun_values(fv), q_point, i)

Ferrite.jl/src/FEValues/CellValues.jl

Line 103 in efa551d

@propagate_inbounds shape_symmetric_gradient(cv::CellValues, q_point::Int, i::Int) = shape_symmetric_gradient(cv.fun_values, q_point, i)

?

[fredrikekre]

I guess this

Ferrite.jl/.github/workflows/Check.yml

Line 17 in efa551d

SKIP: runic

should be SKIP: runic,runic-md now but it is weird that it fails.

[KnutAM]

but it is weird that it fails.

I deleted the cache: "pre-commit-3||6cd8084397da75e58f6d6387e234fe470a36bdbe740bae6dd0cce921aee96fb8", and then it work since it forces re-install of Runic.

I think this errors surfaces because of the update in #1314, as julia-actions/cache v3 now don't delete the cache on failed jobs, so earlier rerunning the job would make this re-install.

Could it be that there really should be an instantiate or smth similar added to the pre-commit workflow @fredrikekre ?