On Thu, May 27, 2010, MengJuei Hsieh wrote:
>
> In the posix standard, the overlapping operations of memcpy and strcpy are
> both undefined. But apparently newer gcc for fedora went a little bit
> further, it crashes.
I thought it would be something like this...thanks for catching that. Do
you think you have caught all the problem strcpy() usages? I was thinking of
substituting our own version of strcpy(), that uses memmove() under the hood.
But that is probably not such a good idea if there are only a few places that
are troublesome.
On another point: Do you want the following commit to be made into a bugfix?
commit a597f08d857589b9e647ac85306ccc94b290ea2d
Author: Mengjuei Hsieh <mjhsieh.gmail.com>
Date: Mon May 24 14:43:54 2010 -0700
1. Corrected the inconsistent default cutres values (PBSA) between
C/Fortran version and the manual. It should be 99 angstrom all
the way. (No test case change is needed.)
2. Fixed an allocation bug in the C interface of NAB PBSA, but this
doesn't solve the problem of recent Intel Compilers on OSX 10.6.
(which is a situation that pbopt values were contaminated.)
Of should we wait for more testing or changes?
...thanks again...dave
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu May 27 2010 - 12:30:03 PDT
