
17 Sep
2008
17 Sep
'08
9 p.m.
Sebastian Pop wrote:
Hi,
[...]
You are right, this is an implementation error in the cloog-ppl backend.
[...]
Yes, the result computed by the PPL is correct.
Sebastian
Sebastian,
if you are interested, I have ready (see attached file) a replacement of function cloog_simplify_domain that uses the PPL implementation: it translates the two cloog input domains into PPL powersets, do the job, and translates back the result into a cloog domain.
Note that in order to use this function you will have to wait for the next PPL snapshot (the last snapshot does not include a few fixes regarding the C interface for powersets). If really impatient, you can try it with the CVS version.
Enea.