[Gretl-users] Bug at Poisson distribution graph.

[Allin Cottrell]

On Sun, 16 Sep 2007, Hélio Guilherme wrote:

> Allin Cottrell wrote:
> > On Sun, 16 Sep 2007, Hélio Guilherme wrote:
> > 
> > > Generating a Poisson distribution graphic breaks because a misnamed
> > > parameter to Gnuplot...
> > 
> > Thanks for spotting that.  Fixed in CVS.

> Sorry, but this fix must be undone. Don't know why but the 
> correct parameter is "lambda1" again.

Sorry, I'm not sure I understand.  

Perhaps this explanation will help: You found a genuine problem 
with the Poisson graphing code, but the fix you suggested was not 
correct: "lambda0" is never a valid parameter name, within the 
context of the graphing code in calculator.c.  My fix in CVS aimed 
to ensure that gnuplot never expected to read "lambda0", but 
always looked for "lambda1" or higher -- and, naturally, to ensure 
that "lambda1" (or higher) is defined in the gnuplot source file.

(The "param1", "param2", etc. business is related to the 
possibility of generating a sequence of distribution graphs -- for 
example, Poisson graphs with lambda = 5, 7, 9, ...  You can do 
this by generating an initial distribution graph then clicking on 
the graph window and selecting "Add another curve".)

Allin Cottrell.