[gmx-developers] Bug in trnio.c ?

[Berk Hess]

>Hi Folks,
>
>I have been modifying some bespoke brownian dynamics code to generate
>gromacs trajectories and have run into some problems with the trn
>header code when reading trajectories.
>
>In do_trnheader (line 110 of trnio.c ) the float precision is
>calculated using sh->natoms. However, in the gromacs-3.2.1 source,
>natoms is not read until after the precision check which causes errors
>from gmxcheck and gmxdump due to the fact that sh->numatoms is
>uninitialised.
>
>Simple fix is to move line 115 ( *bOK=*bOK && do_int( sh->natoms ); )
>to line 105 following the force vector read.

A bug indeed.
But I think Gromacs never writes trn files without box and thus the problem
will not show up with trn files written by Gromacs.

Berk.

_________________________________________________________________
Talk with your online friends with MSN Messenger http://messenger.msn.nl/