Another thing to add is that keoe release works fine. Has anyone else got
a CVS trunk 32-bit linux build to test these files with?
It doesn't look like ARY has had any non-keoe activity recently though.
Tim
On Wed, 25 Oct 2006, Malcolm J. Currie wrote:
>> and given that IACB is reasonable (a 2 or 3) and that ACB_NDIM is a global
>> of 4096 slots or something, it's not clear how that line can segv.
>> Valgrind shows nothing. I haven't tried valgrind on 64bit linux though.
>
> Could something else have trampled where it ought not? Wouldn't
> valgrind reveal that?
>
> The only unusual thing that I noticed in the build is the following
> warning.
>
> ary1_init.f: In block-data unit `ary1_init':
> ARY_ACB:173: warning:
> COMMON /ARY1_ACB1/ ACB_ACC, ACB_BAD, ACB_CHK, ACB_CUT, ACB_DTWEX,
> ^
> Initialization of large (802820-unit) aggregate area `ary1_acb1' at (^)
> slow and takes lots of memory during g77 compile
>
> Looks like one for David.
>
> Malcolm
>
--
Tim Jenness
JAC software
http://www.jach.hawaii.edu/~timj
|