Re: [IMP-dev] More patches with a couple API changes
To: List for IMP development <>
Subject: Re: [IMP-dev] More patches with a couple API changes
From: Ben Webb <>
Date: Fri, 08 Aug 2008 13:19:54 -0700
Reply-to: List for IMP development <>
Daniel Russel wrote:
- The constructors for the nonbonded lists no longer take the radius key
as an argument. They default to XYZRDecorator::get_radius_key and the
value can be set using the set_radius_key method
What if I wanted a particle to have more than one radius? It seems like
this would make such a thing impossible. I can certainly imagine spheres
with a 'hard' inner radius and a 'soft' outer radius, to allow partial
penetration. And all atoms in Modeller have two radii.
- More changes to the names of the hierarchy and molecular hierarchy
decorator functions to make them more consistent. Those should be stable
now
Is there a reason why these are not static methods? Seems like that
would make more sense, and give a cleaner namespace, e.g. rather than
IMP::hierarchy_get_by_attributes(), use instead
IMP::HierarchyDecorator::get_by_attributes().
- I added a function in log.h to control whether exception messages are
dumped to standard error in addition to being embedded in the exception.
The current default is yes. The python lib init code should probably set
it to false for python, but I don't know how to do that.
I can handle that - it's just a %init line in the SWIG wrapper. I'm not
sure why it should ever be true, actually. Surely if I throw an
exception and then catch it somewhere else in the code that exception is
expected, and I shouldn't bother the user with it. Seems like only
unhandled exceptions should be logged.
Other changes look fine to me, and will continue to work their way
gradually into SVN when I have time to disentangle the patch.
Ben
--
http://salilab.org/~ben/
"It is a capital mistake to theorize before one has data."
- Sir Arthur Conan Doyle