algebra::get_unit_bounding_box_d() currently returns a bounding box with
diameter 1 (that is, one reaching from 0 to 1 in each coordinate). This
isn't documented and is kind of inconsistent with
algebra::get_unit_sphere_d() which returns a sphere with radius 1. I'd like
to change the former to be consistent (so
get_bounding_box(get_unit_sphere_d()) == get_unit_bounding_box_d()). It
doesn't look like anything in IMP depends on the old behavior, but it is
used in various tests and examples (at …
[View More]least some of which seem to assume
the behavior I'd like to change it to). Thoughts? Objections?
[View Less]
I'm making some breaking changes to RMF files as part of a cleanup. I think they are all in obscure areas that no one is using yet, but I just want to make sure that no one is surprised:
- bonds will be stored using a different scheme. As a result, existing bonds in the file will be ignored. Adding backward compatibility support for this would be pretty easy.
- the API for storing/extracting bonds has changed (the old API is still there, just no longer documented and now deprecated). There is …
[View More]now a general purpose API for storing data related to tuples.
- the way triangulated surfaces are store is changing (and triangulated surfaces from IMP aren't supported at the moment as I have not gotten around to updating it). It now uses the variable length entries rather than external hdf5 data sets
- the way segments and cylinders are stored has changed
On the plus side:
- you can now associate arbitrary data with any tuple of hierarchy nodes (at the moment, the tuple must have size <= 4). So you can store information about bonds, torsion angles, angles etc.
- mesh surfaces that vary from frame to frame are now supported
- external HDF5 data sets are not used
- polylines and poly-cylinders are now supported
[View Less]