Daniel Russel

Write all IMP particle attributes to RMF file

1

Use the hierarchy to find particles, but other than rigid bodies, don't translate.

Currently they compare the actual list, which can be expensive. Sorting on particleid and then using a hash would be reliable enough and much faster, I think.

Make core::ExcludeBody use grids for rigid bodies

2

Can implement it with algebra::Grid

That is, have IMP particles being RMF Nodes. This is a lot of work, and not clear it is worth it. But would be nice intellectually (as well as remove...

Make it look like other aspects of statistics.

Investigate getting good interrupt support in native code

2

When in C++ code called from python, interrupt doesn't work. This is annoying, and really bad with Ipython.

Run clang-modernize on IMP once we can turn on c++-11

1

It is awesome :-)

Add general CUDA support to cmake IMP

4

- Support .cu files in src. - add IMP_HAS_CUDA macro. - add needed workaround files somewhere. They do not need to be installed, so can just be in the include...

Add move support to Pointer

1

That should make return vectors of ref counted things much faster. http://www.boost.org/doc/libs/1_54_0/doc/html/move/implementing_movable_classes.html

Consistently search in path relative to IMP executable/.pylib to find data

1

To move closer to allowing executables to be run without setup_environment.py as that doesn't play nicely with various ways you want to run things (eg xcode).