Hello,

only to complete the thread...

...I have switched from one suspicious
function that uses lists to one that uses Maps,
but this was not brininging much speed-up.
So my list-based implementation seems to be not that bad.

Then I switched from a string appending function
that uses "^" to the Buffer-module, because I remembered
that it helped my in other programs also to speed up.

I have now about 1/3 less running time for the tool,
and I have not all occurences of the "^" using function
weeped out. So this will bring even more speedup,
when making theses changes complete. :-)

But the general behavour that in the middle of the run
there is a slow-down did not go away so far. gprof also
shows similar percentage values. But possibly the remaining
"^"-applications stops somewhere and somehow, when some very long
strings are concatenated? I will see, when throwing out that stuff
completely.

Ciao,
   Oliver