Subject: Re: [IMP-dev] Iterating over all restraints
From: Daniel Russel <>
Date: Sat, 23 Feb 2008 17:39:42 -0800
Cc:
So in C++ you can just do something like
template <class C>
void visit_restraints(C *m)
for (typename C::RestraintIterator rit= m->restraints_begin();
rit != m->restraints_end(); ++rit) {
RestraintSet *rs= dynamic_cast<RestraintSet>(*rit);
if (rs) {
visit_restraints(rs);
} else {
do something to the restraint
}
}
}
In python I would imagine you can do something similar, but I don't
really know how. Ben?
I don't like the idea of a Flat method (first, because the name is all
wrong). More importantly, in most cases, whether a restraint contains
others (like Bret's do) is really just an implementation detail and is
something that should not be exposed.
What are you trying to accomplish? If it is getting a dependency
graph, you will have all sorts of problems with nonbonded lists and
other sorts of indirect relationships.
How about requiring that Restraint::get_particles() always returns all
the particles that affect the Restraint? It would be horribly
expensive in the worst case, but....
We may need to introduce some lower level mechanism. For example, a
method on Object which returns a list of all pointers contained to
other objects. I may be able to implement that in a clean way with
only the overhead of one extra int per pointer.
On Feb 23, 2008, at 5:19 PM, Keren Lasker wrote:
hi,
I would like to iterate over all restraints.
To my understating the way to do that today is using
model.get_restraints(), although the restraints I get might also be
of type RestraintSet.
It seems that today one basically needs to know which restraint sets
the
model contains.
I suggest to add a function Flat() in class Restraint that returns a
flat list of all restraint in that instance ( will be called