Patch 1804820 : FIT_MAXITER<0

[Philipp K. J. <ja...@ie...>]

Ethan has asked me to take a look at this
patch.

I guess the purpose of this patch is to allow
the user to enter a negative value for 
MAX_ITER, whereupon the fit command will
calculate the residual using the starting
values of all fit parameters and exits. (That is 
really zero iteration steps, but MAX_ITER==0
has already been used to indicate unlimited
iteration).

I am not very happy with this patch:
- I don't understand the purpose. Why do
	you care about the unfitted residual?
- It leads to partially garbage output (the
	errors in the fit parameters are not
	defined for no iteration, etc)
- We are (ab-)using the MAX_ITER parameter
	by introducing a magic value with 
	non-trivial semantics (which is never
	a good idea in my mind).

How does everybody else feel about this?
Who would find this patch useful, and why?

(If it is useful, can we find a better way of
achieving the same goal, rather than shoe-horning
semantics into a negative iteration count?)

Best,

		Ph.