| Thread Previous • Date Previous • Date Next • Thread Next |
Garth N. Wells wrote:
Ola Skavhaug wrote:On Wed, Aug 19, 2009 at 10:14 PM, <kent-and@xxxxxxxxx> wrote:I didn't realise until recently that storage in std::vector is guaranteed by the C++ standard to be contiguous (versus only being contiguous in all known implementations). Using this, I think that we can have safer and cleaner code by using more std::vector in the linear algebra interface. For example, we could have GenericVector::get(std::vector<double>& ); GenericVector::set(const std::vector<double>& ); instead of GenericVector::get(double* x); GenericVector::set(const double* x); For backends that want a pointer to an array, we can pass &x[0]. Any opinions?Is the use of &x[0] safe ? Will it change if the vector is resized ? This is maybe nothing to worry about ?Sure, this is something to worry about, and there will be no magic here :) However, it will enable us to make range checks and issue error messages more easily.We wouldn't let any objects store a pointer to &x[0]. Just as we don't now - it's just data in, data out. On top of the points made by Ola, it will also reduce the likelihood of memory leaks. We can also eliminate some redundant copying of std::vectors into arrays and back.Garth
I've created Blueprint for this: https://blueprints.launchpad.net/dolfin/+spec/stl-vectors-la and set the status to 'Discussion'. Garth
OlaKentGarth _______________________________________________ DOLFIN-dev mailing list DOLFIN-dev@xxxxxxxxxx http://www.fenics.org/mailman/listinfo/dolfin-dev_______________________________________________ DOLFIN-dev mailing list DOLFIN-dev@xxxxxxxxxx http://www.fenics.org/mailman/listinfo/dolfin-dev_______________________________________________ DOLFIN-dev mailing list DOLFIN-dev@xxxxxxxxxx http://www.fenics.org/mailman/listinfo/dolfin-dev
| Thread Previous • Date Previous • Date Next • Thread Next |
