VMD-L Mailing List

From: John Stone (johns_at_ks.uiuc.edu)
Date: Sat Apr 19 2008 - 14:29:46 CDT

Hi,
  In practice all of the warnings emitted with -Wall
when compiling VMD are of no consequence. A large fraction
of them are complaining about byte swapping functions that we
declare statically in a header file becuase each plugin must not
have any external dependencies other than the math library and such.
All of these warnings about unused byte swapping routines can be completely
ignored. We enable -Wall for compiling VMD for our own purposes. If
people find these warnings disturbing, they may with to compile without
it. None of the warnings that are presently emitted are anything to be
concerned about. Even the warning about the type punned pointer
can be ignored, as there should be no possibility of aliasing of paramters
in the code.

Cheers,
  JohnStone

On Sat, Apr 19, 2008 at 10:21:04AM -0400, Axel Kohlmeyer wrote:
> On Sat, 19 Apr 2008, Cojocaru,Vlad wrote:
>
> VC> Dear VMD users,
>
> dear vlad,
>
> first off, it would make your mail much more readable for many
> people, if you restrict it to plain ASCII characters...
>
> VC> I am building MD from source (CVS) on a LINUXAMD64 (SuSE linux, gcc
> VC> 4.2). During the plugin build I get a lot of warning messages. Some
> VC> examples are below. The build seems ok inspite of these warnings. My
> VC> question is: Shall I worry about these warnings .. or I can go on
>
> warnings always should be taken seriously. the default VMD builds,
> however are done at a high warning level, most of the resulting
> warnings are rather harmless. with few exceptions they _do_ indicate
> not fully standard conforming code or sloppily written code, so
> the could should be corrected at some point. but you are obviously
> building the plugins and most of that code is contributed, so it
> is on average in a worse shape than the main part of VMD.
>
> for most warnings it is rather obvious what to do to get rid of them.
> feel free to do it and send in a patch so that everyone benefits from it.
>
>
> VC> with the build ?
>
> [...]
> VC> src/msmsplugin.C:201: warning: dereferencing type-punned pointer will break strict-aliasing rules
>
> this is the only one that may indicate a more serious problem.
> recent gcc compilers assume full compliance of the code with ANSI-C
> aliasing rules, but compiler has found a region where this seems
> not to be the case. this may result in the compiler miscompiling
> this plugin and can be alleviated by either fixing the code or
> compiling it with -fno-strict-aliasing which will relax the aliasing
> constraints at the expense of skipping some possible optimizations.
>
> cheers,
> axel.
>
> VC>
>
> --
> =======================================================================
> Axel Kohlmeyer akohlmey_at_cmm.chem.upenn.edu http://www.cmm.upenn.edu
> Center for Molecular Modeling -- University of Pennsylvania
> Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
> tel: 1-215-898-1582, fax: 1-215-573-6233, office-tel: 1-215-898-5425
> =======================================================================
> If you make something idiot-proof, the universe creates a better idiot. 

-- 
NIH Resource for Macromolecular Modeling and Bioinformatics
Beckman Institute for Advanced Science and Technology
University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
Email: johns_at_ks.uiuc.edu                 Phone: 217-244-3349
  WWW: http://www.ks.uiuc.edu/~johns/      Fax: 217-244-6078