Re: "Too many arguments", SuSE 6.2

[Matt Aylward]

On Mon, 29 Nov 1999, you wrote:
Marcus G. Daniels: >>>>>> "MA" == Matt Aylward <address@hidden> writes:
Marcus G. Daniels: >
Marcus G. Daniels: >MA> The backtrace goes on for ever with the search 
(localSearchCell)
Marcus G. Daniels: >
Marcus G. Daniels: >If it really does go on and on and on, then you could be 
running out
Marcus G. Daniels: >of stack (typically 8MB).  It would seem to depend what the 
value of
Marcus G. Daniels: >`CORE' is, and the relationship between `lattice' and 
`input', i.e. whether
Marcus G. Daniels: >or not it was possible for endless oscillation back and 
forth to
Marcus G. Daniels: >occur.  

CORE is a #defined constant = 4
Looking back at this, it is pretty dangerous stuff. 'lattice' and 'input'
refer to the same Discrete2d. The 'input' contains values of 0 or 1, the
search marks every cell containing a '1' with a 4 (CORE) if it's patch is
above a minimum area, and 3 (LONER) if the patch is smaller, so the search
should eventually meet it's own starting point which was labelled CORE at
step 1 (i.e.,
"if ((int) *discrete2dSiteAt (lattice, offsets, xLocn, yLocn) == 1)"
fails for the whole neighbourhood and the search follows itself back filling
in any gaps. It used to work beautifully but I guess it really should be
redesigned.
If the biggest Discrete2d was a single patch, it is 180x180 = 32400 cells.
It used to work in that case. 
It now appears to be segfaulting at less than 1000 cells in the biggest patch.

Marcus G. Daniels: >Also, I guess it was just a typo that there was no
Marcus G. Daniels: space after >"yLocn"?

Yes a typo, it is not in the code.

I will look closer and redesign it to avoid the recursion.
Thanks again,
Matt

--
      Matt Aylward <address@hidden>
Deptartment of Botany, University of Western Australia
   Long live Linux, Swarm modelling and Grass GIS.

                  ==================================
   Swarm-Support is for discussion of the technical details of the day
   to day usage of Swarm.  For list administration needs (esp.
   [un]subscribing), please send a message to <address@hidden>
   with "help" in the body of the message.