The star formation/supernova feedback implemenation was translated
from Gasoline by Greg Stinson.  Note the following differences from
the Gasoline implementation that has been in common use from 2006
through 2010.  N.B. In all of the following, the definition of
"metals", "Oxygen" and "Iron" are not clear, and this may be the
origin of some of the noted differences.  E.g., does "Oxygen" refer to
the mass of O16 in particular or the mass of all the alpha elements.

* Upon reading a tipsy file which only has "metals",  the division of metals
  between "Oxygen", and "Iron" is 95% and 5% respectively, whereas in Gasoline
  this division was 75% and 25%.  The 75%/25% in gasoline was a rather random
  guess.  95%/5% is also incorrect.  The real answer lies somewhere inbetween.
  It seems sensible to put the oxygen/iron abundances to the solar value.
  If you follow this stuff, that ratio has changed dramatically over the last
  10 years from 0.84/0.12 from Anders + Grevesse (1989) to 0.58/0.13 in 
  Asplund et al. (2005?).  See the review by Asplund et al. (2009).  The
  trouble is that our SNII yields are based on the former,
  though it might be nice to use the latter.  So, Oxygen should be somewhere
  between 81% and 87%.

* SNIa eject all their mass in "metals", whereas Gasoline only ejects .76
  solar masses in "metals" per supernova.  SNIa models are of exploding
  Oxygen white dwarfs, so all the material in such an explosion is composed
  of "metals" (Oxygen or heavier elements).  In Gasoline, Fe + O =0.63+0.13
  solar masses = 0.76.  Nomoto et al (1999) Table 1 states that Fe is
  actually 0.74 once you count up all of its isotopes.  Still, the rest of
  the ejecta are metals, lots of Si and Ni, so this will contribute to metal
  cooling.
  
* The minimum mass of binary that can go supernova is .8 solar masses, whereas
  in Gasoline this is 3 solar masses.  The lowered limit allows SNIa's to
  continue happening for a very long time after star formation.

* In Gasoline, mass lost in SNIa is discounted against mass lost in winds.
  With the lower limit to SNIa binary masses noted above, which is below the
  lower mass limit for winds, there can be SNIa going off and no winds.
  Discounting SNIa mass loss against the wind mass loss would result in
  negative mass loss in winds, so it is not done.