dtb.sac slays the entire planet again, but with gusto!
|
|
Bugzilla Link |
267 |
Created on |
Aug 09, 2006 23:19 |
Resolution |
FIXED |
Resolved on |
Aug 11, 2006 10:32 |
Version |
1.00beta |
OS |
Linux |
Architecture |
PC |
Attachments |
crud2.sac, crud3.sac
|
Extended Description
The attached is a dinkytoy version of dtb.sac with lotsa stuff
ripped out of it, in the interest of keeping Der Bodo in good spirits.
It has the following set of interesting behaviors on build 14875
(with an rbe-tinkered version of shape_cliques.c that I think it not
relevant to this...):
sac2c -noopt crud2.sac
- works OK.
sac2c -O3 -noive -nosci -noisv crud2.sac
- reduce gives proper result
- contemplates navel, then runs out of memory
sac2c -O3 -noive -nosci
- result shape for reduction wrong
- runs out of memory as above
I tried foldfix vs nofoldfix, but that appears to be innocent.
There was also some funnies happening with strange shape vectors
with zeros in them where, like the hat on the bedpost, shouldn't have
been there.
There was also something where a Boolean array print had a value of
5, but I'm getting cross-eyed now...