Commit summary from repository starlink
------------------------------------

  (Ed Chapin) smurf: don't segv on un-initialized smfFilter
   100.0% applications/smurf/libsmf/
  
  (Ed Chapin) smurf: check for existence of filter only with real part.
   100.0% applications/smurf/libsmf/
  
  (Ed Chapin) smurf: don't forget to initialize flt.whiten
   100.0% applications/smurf/defaults/
  
  (Ed Chapin) smurf: initial commit of whitening filter
    95.5% applications/smurf/libsmf/
     3.4% applications/smurf/libsmurf/
  
  (David Berry) prm: Add prm_par.h constants giving the HDS code for each data type
   100.0% libraries/prm/
  
  (David Berry) kappa: Correct ndfcompress docs
    48.7% applications/kappa/libndfpack/
    51.2% applications/kappa/
 

Commits from repository starlink
-----------------------------

  commit cbbe0c241f163a60879ab9f8f44072e055193787
  Author: Ed Chapin <[log in to unmask]>
  Date:   Tue Oct 12 17:10:05 2010 -0700
  
      smurf: don't segv on un-initialized smfFilter
      
      This can happen if you set flt.whiten. It fails miserably because
      the filter itself maybe uninitialized and we're just trying to
      call smf_whiten iteratively. Not sure of the best way to solve
      this problem at the moment, but we should stay away from
      flt.whiten... whitening can only be used as a pre-processing
      step at the moment.
  
   applications/smurf/libsmf/smf_filter_complement.c |    3 +++
   1 files changed, 3 insertions(+), 0 deletions(-)
  
  commit f6b119ef4bf57fda18aa161f24278500c32ced39
  Author: Ed Chapin <[log in to unmask]>
  Date:   Tue Oct 12 16:56:45 2010 -0700
  
      smurf: check for existence of filter only with real part.
      
      Unless we're dealing with a complex filter (which we never do at
      the moment), the imaginary part will usually be NULL anyways. Only
      need to check for the existence of the real part of the filter.
  
   applications/smurf/libsmf/smf_filter_execute.c |    5 ++---
   1 files changed, 2 insertions(+), 3 deletions(-)
  
  commit 0809db84af6c634bcb575972e45ba11a7ad83bf5
  Author: Ed Chapin <[log in to unmask]>
  Date:   Tue Oct 12 16:56:23 2010 -0700
  
      smurf: don't forget to initialize flt.whiten
  
   applications/smurf/defaults/smurf_makemap.def |    1 +
   1 files changed, 1 insertions(+), 0 deletions(-)
  
  commit d90b8b84a355ab2b464adf2f5960f44794c11831
  Author: Ed Chapin <[log in to unmask]>
  Date:   Tue Oct 12 16:30:30 2010 -0700
  
      smurf: initial commit of whitening filter
      
      smf_whiten fits a power-law to the low-frequency end of the data,
      adds it to a flat value that is fit to a higher-frequency portion
      of the spectrum to estimate the white noise level, and then divides
      by this function to create a nearly flat power spectrum.
      
      It is currently called within smf_filter_exectute. There is a
      new configuration parameter called "whiten" which, when set, will
      cause whitening to be applied as a pre-processing step (it is
      therefore also available to sc2clean). smf_filter_execute was
      modified slightly so that if an empty filter is supplied, it
      will still run and only apply the whitening factors.
      
      To see what it does, use sc2concat to combine some data, run
      it through sc2clean with default parameters, and a second
      time setting "whiten=1". The latter will have essentially no
      large baseline drifts. You can also run both files through
      sc2fft with "power=true" set, and you will see how the
      low-frequency end has been flattened out. The only remaining
      features are lines in the power spectrum (some narrow, and
      some quite wide), as well as the anti-aliasing filter which
      begins to noticeably roll-off the power spectra above about
      50 Hz.
  
   applications/smurf/defaults/smurf_sc2clean.def    |    2 +
   applications/smurf/examples/dimmconfig.lis        |    7 +-
   applications/smurf/libsmf/Makefile.am             |    1 +
   applications/smurf/libsmf/smf.h.source            |   11 +-
   applications/smurf/libsmf/smf_calcmodel_flt.c     |    2 +-
   applications/smurf/libsmf/smf_calcmodel_noi.c     |    2 +-
   applications/smurf/libsmf/smf_checkmem_dimm.c     |    2 +-
   applications/smurf/libsmf/smf_clean_smfArray.c    |   16 +-
   applications/smurf/libsmf/smf_filter_execute.c    |   66 +++--
   applications/smurf/libsmf/smf_filter_fromkeymap.c |    3 +-
   applications/smurf/libsmf/smf_get_cleanpar.c      |   20 +-
   applications/smurf/libsmf/smf_get_padding.c       |    2 +-
   applications/smurf/libsmf/smf_whiten.c            |  291 +++++++++++++++++++
   applications/smurf/libsmurf/smurf_sc2clean.c      |    7 +-
   applications/smurf/libsmurf/smurf_sc2threadtest.c |    4 +-
   15 files changed, 390 insertions(+), 46 deletions(-)
  
  commit 51e3a4a5fac5bca4edbd2e5e61d2acd8013ee802
  Author: David Berry <[log in to unmask]>
  Date:   Tue Oct 12 11:03:32 2010 +0100
  
      prm: Add prm_par.h constants giving the HDS code for each data type
      
      The new constants are called VAL__NAM<X> and have values "_DOUBLE",
      "_WORD", etc.
  
   libraries/prm/make-prm-par.c |   29 ++++++++++++++++++++++++++++-
   1 files changed, 28 insertions(+), 1 deletions(-)
  
  commit 6710dc79fa32149f03944ed452485e6a55ca7f40
  Author: David Berry <[log in to unmask]>
  Date:   Tue Oct 12 09:29:56 2010 +0100
  
      kappa: Correct ndfcompress docs
      
      The documented parameter name "SCALETYPE" was wrong. The parameter name
      actually used is "SCALEDTYPE".
  
   applications/kappa/libndfpack/ndfcompress.f |   15 +++++++++------
   applications/kappa/sun95.tex                |   12 ++++++------
   2 files changed, 15 insertions(+), 12 deletions(-)