Attendance:  
----------

Duncan, Patrick, Warren, Joe, John, Greg, Siong, Kent, Masha, Phil



WrapperDatatypes:  
----------------

Changes are into LALWrapper and ready to go.  Masha put in
changes including base-frequency, etc into the frequency series.  Kent was not
sure why this had been done at this time since the metadata for frequency
series has not been ratified by the SCCB.  The real point of this change was
to put some information which was being treated as history during the past MDC
into datatypes.   The decision was made to let it stand as is,  and wait for
ratification of the conventions document for further changes since there may
be some changes.   Kent asked how these variables would be populated.  The
datacond is not ready to read in this information in ILWD yet,  but it will be
worked on.   



Signals and other stories:
-------------------------

Julien's code had a memory leak  so that a signal was being raised
in LALFinalizeSearch().  MPIapi did not see it because of order of mpiAPI
shutdown versus finalization on wrapperAPI.  According to Jolien,  there 
is a possibility that there is a bug that can cause things to hang if the 
heap is corrupted but this seems unlikely.   Much more likely that there is
some problem with order of finalizations in wrapperAPI and mpiAPI.     

Since the original e-mail circulated by Julien to Jolien and Alan (??)
mentioned threads,  there was a discussion of threads and thread-safe code.
Kent,  once again, emphasized that the MPI libraries are not thread safe.  The
developers at Argon are re-writing core of MPI so that it can be thread safe.
It's technically difficult.  There is no reason to be thread safe;   but
making LAL and LALWrapper thread safe might be a good thing.  LALWrapper 
is not thread safe since there is a global variable which is visible in 
shared objects.   There is no way around this because this variable is a
void* used to carry search specific information.     



Conventions again:
-----------------

Discussion of the power spectrum normalization came up again.  Along with a
change from one sided to two sided,  the issue of units needs to be addressed.
How are the units in the ILWD handled?  How is the units field of a PSD set
What is the method used by frames?   Rather than re-invent the wheel,  Kent
will talk to Alex and John will look at the frame-library and spec.  Generally
agreed that method used in frames should stand unless it is grossly wrong.  
Decided to postpone discussion of conventions document until next week.  



Tasks:
------

Masha will check out and build latest LAL and LALWrapper at Caltech.  Once its
done,  she'll send an e-mail to the group.

Duncan and Masha will confirm the suspicion about the signal by getting and
running Julien's old code that caused the problem during the MDC.   Duncan
will check the 05sick tests from the MPI MDC in standalone mode:  if they are
handled in standalone,  then will test it within LDAS.