Jeremy
Jeremy
@gvoskuilen We should dump a `component-info.csv` or similar giving potentially more information about each component for use in post-processing statistics. In post-processing, you could then do a table join on...
Newer versions of xcode will no longer have a /usr/include in the path, meaning many macport/homebrew/default compilers will NOT bring basic library headers into the include path by default. The...
We currently only support output of integer/float types. There are certain scenarios where it would be useful to output strings (extra metadata fields). This is already done to a certain...
ComponentInfo just points to objects inside ConfigComponent, which means ComponentInfo is pointing to deleted memory. The main problems are Params and stat enable list. Params and EnableLists should really be...
This is a general cleanup of memory leaked at the end of the simulation so that valigrind/leak sanitizer are happy
GCC < 4.8.5 will accept the -std=c++11 flag, confusing the configure into thinking the compiler supports all of C++11. Previous versions of gcc support only a subset. Configure happily finishes,...
During config graph construction, zero-latency links should not be allowed. Error (or maybe warning) should be generated. Lack of error check leads to seeming deadlock in the application as it...
Code needed for serialize_serializable: ```` template class serialize { public: void operator()(std::shared_ptr& o, serializer& ser){ T* tmp = o.get(); serialize_intrusive_ptr(tmp, ser); o = std::shared_ptr(tmp); } }; ```` Also, should probably...
It would be good to have clang builds of sst-core/sst-elements tested. Additionally, it would be good to have address sanitizer builds tested for better/more rigorous memory error checks. I have...
## Bug Report @trilinos/tpetra ### Description The unit tests for Tpetra are failing to build with undefined references to `endFill` for the `FECrsMatrix` and `FECrsGraph` classes. ### Steps to Reproduce...