\begin{verbatim}
CHANGELOG file
last updated 04 Feb 2003
+------------+
|version 1.01|
+------------+---------------------------------------------------------
image1D.f:
----------
1) fixed the bug in calling CERNLIB that kept knotmode 3 from working under 
linux
2) more comments

configure/build:
----------------
added scripts to the config directory since not all systems get shipped 
with them

+-----------+
|version 1.0|
+-----------+----------------------------------------------------------
Overall:
--------
1) new autoconf/automake configuration and build scheme
2) new control scripts (hopefully more rational and more portable)
3) new directory structure
4) removed 3d codes.  save those for version 2.0
5) finished documentation

makeco1D.f:
-----------
1) fixed option 100 (load from disk)
2) fixed case when data has un-equally spaced bins

genmaPP1D.f:
------------
fixed Coulomb bug introduced sometime in last few months

+-----------+
|version 0.9|
+-----------+----------------------------------------------------------
Overall:
--------
1) needlessly inflated the version number
2) rewrote setup script in python

rawKspline.f:
-------------
1) unified all kernel loaders in one routine
2) unified all kernel functions in one routine
3) switched to higher order polynomial for interpolating kernel table
4) added meta-data to kernel files

+--------------+
|version 0.09.3|
+--------------+-------------------------------------------------------
In image1D:
-----------
errors were computed incorrectly -- it is now fixed and the errors on
S(r), integral of S(r), and restored correlation are much smaller now.

In lib/:
--------
random number generator changed so that it generates a different sequence 
of random numbers each time it is called

+--------------+
|version 0.09.2|
+--------------+-------------------------------------------------------
In the genmaXX files:
---------------------
1) renamed them from genmaXX -> genmaXX1D because there are also
genmaXX3D kernels in the works.  The genma_meson.f code seems
to be an exception to the naming scheme.  Well, it is capable 
of producing 1D *and* 3D kernels for any mesons (you have to 
change particle.inc however)
2) the format for saving the kernel is different.  now the format is 
formatted and it specifically is:
   NSQ NSR
   K_0(1,1) K_0(1,2) K_0(1,3)....
   K_0(2,1) ...
   .
   .
   .
   l=0
   K_1(1,1) K_1(1,2) ....
   .
   .
   .
   l=1
   ...
here, NSR and NSQ are the number of steps in the r and q directions.
K_l is the lth componant of the kernel (expanded in legendre polynomials
for 3D work). 
3) there are alot more bins in r and q.  also, the way i have it rigged
now, you don't have to rebuild the kernel each time you change the binning
in q
4) no longer used interpolation to speed up the potentials.  since the
kernel doesn't have to be remade so often, it can take longer to build

In makco:
---------
1) makco has more choices
2) more accurate because kernel better 

In the image1D:
---------------
1) b-splines
2) equality constraints
3) checks inequality constraints
4) optimized knots found using faster/better minimization routine
5) can output the optimal resolution matrix (some day i'll have to explain
what this is)

Other stuff:
------------
1) is a 3D version of image1D (it was in the first version of the new code
i put on the web, but i took it out of the last one as it's not really
done testing yet)
2) viewsourceXD codes, so you can remake the images if you only have the
coefficients and knots.
3) mtx2xpm, a little c++ code to make any of the matrices made by the
codes into a grey-scale pixmaps
4) fitsource1D, a little code that fits gaussians to imaged sources and a
script fitit to automate the process
5) genmaDA1D and smearker (thanks Giuseppe!)

Overall:
--------
1) centralized controls whereever possible (e.g. main.controls)
2) fewer include files to change and they are named more intelligently.
3) using CERNLIB whereever possible (it's reasonably fast and available
everywhere)
4) runs on more platforms:  wrote it on Linux (egs f77) and on OSF (dec
fortran)
5) scripts for automating some boring things and for giving more
reasonable filenames to things


+------------+
|version 0.03|
+------------+-------------------------------------------------------
1) added option of inputing source from disk into makco.f
   (this allows you to build model sources using other programs like RQMD)
2) simplified Makefile
3) fixed linking bug in installib.com (thanks to Dennis Reichhold 4/7/99)

+------------+
|version 0.02|
+------------+-------------------------------------------------------
1) minor bug on some systems in lib/simplx.f
2) minor fixes in makefile that allow programs to run under Linux
3) added Woods-Saxon like source to makco.f
4) added realistic noise to sources from makco.f
5) fixed bug that made standard deviation of noise in randomnoise sqrt(2)
   too big
+------------+
|version 0.01|
+------------+-------------------------------------------------------
original version
\end{verbatim}