Attendance:  Patrick, Kent, Masha, Phil, Joe, Warren, John, Greg, Steve
Berukhoff,  Curt Cutler,  Duncan,  Jolien,  Alan.  



Ephemeris data for periodic searches:
------------------------------------

Curt summarized the resource files need to compute the Doppler corrections in
searches for periodic sources.   These files computed using the JPL emphemeris
consist of a 9 dimensional array sampled once per hour.  A year of data is
about 1/2 a Mbyte.  This format is supported by the multidimdata in the
wrapperAPI and clearly fits into an existing ILWD container.   Once this was
clear,  Kent proposed that these data should be cast into ILWD format for
ingestion by the datacondAPI and susequent transmission to the wrapperAPI
using the intermediate command.   After some discussion,  it was agreed that
the data should be part of the LAL source tree as ascii files (residing in the
test subdirectory of Curt's pulsar package).   At build time,  these files
would be parsed into ILWD format and installed in $PREFIX/share/lal.   Kent
mentioned that it would be best to have these in compressed ILWD format so
that they can be read more quickly from disk.   The parsing into ILWD would
optimally involve adding a new utility program to the LAL source tree which
would only be built if a new configure flag --enable-ilwd were used since it
will probably have to link against the ilwd.so library in LDAS.   



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

Duncan and Masha tracked down the problems associated with job termination in
the wrapperAPI and mpiAPI if a memory leak is detected by lalwrapper.   It
turned out that there were a couple of minor bugs in the wrapperAPI source;
they have been fixed now.   Duncan submitted a PR to the LDAS bug-tracking
system to maintain an audit trail.  



LDAS,  LALWrapper and LAL:
-------------------------

Kent mentioned that there were some problems with the compiler when the last
release of LDAS was made and the versions of LDAS at remote locations were not
up to date.   LDAS has switched to a newer version of libtool which has broken
LALWrapper and LAL.   This is being investigated actively by Jolien;  as an
interim solution one can pass a --disable-dependency-checking at configure
time to force the build to complete.  Jolien will e-mail the appropriate lists
about this.

Stochastic DSO:  Outstanding issues.  LAL does not have double precision versions
of some things.  Bar people seem to need higher precision to deal with some of
the operations they need to perform.   This showed up during the MDC when the
bar stochastic group were comparing the output of Siongs shared object with 
output of Matlab code.  They found there was disagreement at about 1/10^3 or 
1/10^4;  it also stepped thus indicating a precision issue.  Indications are
that this may be arising from a LAL routine to average some power spectra.
John will look into it by re-writing that routine in double precision and
doing further comparison.  There is some concern that this may need a double
precision FFT routine,  but we will play a wait and see game here.    



LDAS Releases and coordination of software development:
------------------------------------------------------

There is a release of LDAS scheduled for the middle of October.  What should
we do about LAL and LALWrapper?   As a general rule,   we will coordinate
releases between these three packages.   But in this case,  it seems that the
spirit of the LDAS release is to fix bugs and should not require a release of
LAL or LALWrapper.   We will determine a schedule for releases during next
weel's meeting after coordination with other groups.   For this release of
LDAS,  we will simply tag LAL and LALWrapper.


Other business:
--------------

LALWrapper HOWTO  will be discussed next week.