20Nov2007: -  Keep a CHANGELOG  ;-)
           -  Keep track of version numbers.  We're now at 0.2

26Nov2007: -  Made some fixes with Mark's help.
	   -  Incorporated Krisciunas' JHK polynomial templates to both
	      dm15temp and dm15temp2 generators.
	   -  sn.plot()  now blanks out bad polynomials

03Dec2007: - Needed to change the 'super' class to 'sn'.  'super' is a
             reserved function in python to obtain a class' super-class,
             which is super-handy.
           - Updated the load() function to test if the saved file is
	     using 'super' and upgrade if needed.
	   - Added the fit_colors functionality to fit() (see docs).
	   - Fixed how get_restbands works (normalization was bad)
	   - set_generator() now automatically calls get_restbands
	   - Major overhaul of mangle_kcorr().  It now recognizes the bad
	     data masks and uses it when computing colors:  the JHK data
	     should now be mangle-able.
	   - Set Hsiao's opt+NIR uberspectrum (H3) as the default for
	     k-corrections.
	   - Updaded default behaviour of fit():  s=1 is now default, set it
	     to None to make it vary.
           - Fixed lc.plot()  to properly show residuals from template fits.

20Jan2008:  pylc is now versioned using SVN
06Jan2009:  pylc is now SNpy (affectionately SNOOPY).
01Feb2009:  Begin the arduouis task of converting over to Numpy, the new
            Scipy, and, heh what the heck, python 2.5.
05Oct2011:  Incorporated the new light-curve templates based on the data 
            released in Stritzinger et al. (2011).  Also, try to remove
            a number of dependencies that are causing problems when porting
            to new architectures.  Sorry, PGPLOT, but you're just too 
            hard to maintain.

Feb2012:
            SNOOPY2 is now released.  New templates using the stretch parameter
            instead of dm15 exclusively.  For now, we still use the dm15 as the
            default parameter, but I think I'll change that at some point...
            PGLOT is now GONE.  I'm sick of trying to port it.  Long live
            matplotlib.

            There is now a new method for installing:  using a bootstrap
            script form virtualenv.  This will allow SNooPy to exist side
            by side with other modules and not interfered.  That's the idea
            anyway.

August 2012:
           SEveral changes and addons.

           - Added the following filters:
             2MASS filters (J2m, H2m, Ks2m)
             UKIRT filters:  WFCAMz, WFCAMY, WFCAMJ, WFCAMH, WFCAMK
             WIYN filters:  WHIRCJ, WHIRCH, WHIRCK
           
           - When adding filters to SNooPy, instead of specifying the zero-point
             you can now simply put in strings like VegaB=0 or BD17=15.0
             The meaning is that this filter has a VegaB (Bohlin CALSPEC Vega SED)
             magnitude of 0 or a BD+17 (again, Bohlin CALSPEC version) magnitude
             of 15.
           
           - When you want to fit a light-curve using one over many different
             interpolators (use list_types() to see a list), you can also
             do it interactively by passing 'interactive=True' to the template
             function.  For example:
             >>> s.B.template(method='hyperspline', interactive=True)
             This will pop-up a plot with residuals.  Type '?' for key
             binding help.  Also see SNooPy documentation section 7.5.3
           
           - Compute bolometric light-curve using the bolometric() function.  This
             is experimental.  the function signature is.  To see how to use it, do:
             >>> s = get_sn(...)
             >>> help(s.bolometric)
           
           - If you want to solve for a single mangling function that is
             the same for all epochs, pass the option through the fit
             function in the following way:
             s.fit(..., margs={'mepoch':True})
             or, if computing the k-corrections manually
             s.kcorr(mangle=1, mepoch=True)
           
