December 2009 - IMP-dev

simple transformation test fails
by Keren Lasker 23 Dec '09

23 Dec '09

hey, can someone please help me figuring out what is the problem with the attached test. I am keep getting an error: UsageException: Attempt to use uninitialized vector. The error seem to come from VectorD, but I really think that all vector are initialized. If I initialize Rotation3D from 4 number everything is ok. ?? thanks keren.

2 1

Make ScoreStates more comprehensible
by Daniel Russel 15 Dec '09

15 Dec '09

A number of people have found score states confusing both because the name doesn't really suggest much useful as well as because it is not obvious that they are the place to look if you want to implement a constraint. To address this several of us came up with a few proposed changes: - rename ScoreState to ModelState. This rename would make the names consistent with OptimizerStates (XStates maintain state associated with X) and get rid of the idea that they have something direct to do with scoring. It also could be done trivially in all user code that uses ScoreStates via "sed -i.old 's#ScoreState#ModelState#g' **/*.h **/*.cpp **/*.py". - Create a new base class Constraint which inherits from ModelState. The new base class wouldn't (necessarily) add any functionality, but it would provide a place in the documentation to discuss "constraints" as defined by the Sali lab and label ModelStates that are constraints as such. We would have to decided whether to rename existing XScoreStates which are constraints to XConstraint. These renames should, I think, be accomplishable through sed and hence not too painful. - Continue with hiding ModelStates from the basic user experience. Currently, one should not have to know what they are and what they do unless you are implementing code for IMP. The documentation doesn't reflect this yet (as it was not originally the case). I'll work to further mark them as advanced. Comments? Thoughts? Opposition?

1 0

imp-dev meeting Tuesday at 1pm
by Daniel Russel 14 Dec '09

14 Dec '09

Since we are trying to clean everything up for the release, I thought it might be good to have a meeting to get people's thoughts on - the IMP documentation: what needs improvement - coming up with a checklist set of standards for documentation to provide some a better starting point than "write some examples and fill in the doxygen docs" - what is a good balance between what I'll call "reference" documentation such as doxygen docs and general examples of using some component and "cookbook" examples which provide a solution to a scenario which is hopefully close to what the user wants (but, on the down side, tends to provide little information about how to solve other problems). Please think about how you use the IMP documentation and documentation of other libraries. What other libraries provide particularly good docs and what guidelines can we draw from those? What are cookbook examples that IMP should include? See you tomorrow.

1 0

particlepair::first
by Daniel Russel 11 Dec '09

11 Dec '09

Is anyone in the habit of using first and second (as opposed to [0] and [1]) to access members of particle pairs?

1 0

Stubborn IMP with boost
by Javier Ángel Velázquez Muriel 02 Dec '09

02 Dec '09

Hi, I am receiving this error for the first time after just updating: Checking for Boost version >= 1.33... Boost version >= 1.33 is required to build IMP, but it could not be found on your system. Please see config.log for more details. In particular, if you have Boost installed in a non-standard location, please use the 'includepath' option to add this location to the search path. For example, a Mac using Boost installed with MacPorts will have the Boost headers in /opt/local/include, so edit (or create) config.py and add the line includepath='/opt/local/include' I'm trying to include it directly with this line in config.py, but no luck: includepath='/usr/include/boost' Help! ;)

2 5