Solve the BV problem by asymptotic matching techniques:

 	ep y'' + 2y' + exp(y) = 0,  y(0)=0, y(1)=0

where ep << 1.   There is a boundary layer as x->0.

Caution:  This is a NL eqn.  When scaling, you will have to carry around 
y's.  To scale then, one must know the "size" of y.  As a start, assume 
that the size of y is of the order expected from the outer solution.  Use 
this to approximate the inner layer.  If in doubt, let y~O(1).  However, 
regardless of your original scaling, check self-consistently that what you 
threw away is smaller than what you kept.  When I do this, I find that 
there is a scale to the boundary layer and that my solution in the BL is 
valid but breaks down when x gets much larger than the BL scale (but not 
overly large).  Identify this second larger scale size.