Don't make constructor of MappingQ explicit, after all. This sounded like a

good idea but at least gcc 4.1 has a bug in which it complains that the
following code can't be compiled:
.....................
struct Mapping {
    Mapping ();
    explicit Mapping (const Mapping &);
};

void push_back (const Mapping &);

int main () {
  push_back (Mapping());
}
.....................
The error is erroneous, but we get hit by it some 20 times in the
testsuite. I'm also not entirely sure whether it's worth the effort of putting
a test for this in aclocal.m4 and making the constructor explicit only if the
test succeeds...

git-svn-id: https://svn.dealii.org/trunk@24167 0785d39b-7218-0410-832d-ea1e28bc413d

diff --git a/deal.II/include/deal.II/fe/mapping_q.h b/deal.II/include/deal.II/fe/mapping_q.h
index c0c996682d7a6773aa30886d1703c5f77d4b0231..50ffaf7eeda0faf27e9b031b4534d02b469f3721 100644 (file)
--- a/deal.II/include/deal.II/fe/mapping_q.h
+++ b/deal.II/include/deal.II/fe/mapping_q.h
@@ -82,7 +82,7 @@ class MappingQ : public MappingQ1<dim,spacedim>
                                      * #tensor_pols pointer as done
                                      * by a default copy constructor.
                                      */
-    explicit MappingQ (const MappingQ<dim,spacedim> &mapping);
+    MappingQ (const MappingQ<dim,spacedim> &mapping);
 
                                     /**
                                      * Destructor.