Re: [AMBER-Developers] test request for ambertools and Snow Leopard 64 bits kernel

From: Volodymyr Babin <>
Date: Mon, 30 Nov 2009 14:27:17 -0500 (EST)

Hi Ben,

> OK, I've given that a go, to no real avail. It still dies in the same
> place. I haven't filled the bondtype.cpp up with debugging statements,
> as I suppose it's better to focus our energies on bondtype.c at present.
> I've attached the output file I got from the C++ version, though it
> doesn't really tell me anything new.

It does say something to me though: both .c and .cpp are broken
[this implies that the problem most likely is in the part that
is shared by the two].

>> As far as I could understand the 'bond2' array gets malloc-ed and not
>> initialized (at least some members do not get initialized); therefore
>> those different values are to be expected (it is just garbage). OTOH,
>> this is not necessarily an error.
> I understand. So much for an obvious and easy solution...

It was remotely not a solution, just explanation of your observations
of very different values (of uninitialized memory) seen on linux and
leopard. I did not go though the loops deeper because: (a) we do not
have leopard here, (b) Junmei should be able to navigate though this
part much faster than me.

With kind regards,


AMBER-Developers mailing list
Received on Mon Nov 30 2009 - 11:30:07 PST
Custom Search