[gmx-users] RE: Mark & Justin - grompp error for gmxtest-4.0.4 and 3.3.3 on new gromacs-4.0.5 install

[Justin A. Lemkul]

Yirdaw, Robel Birru wrote:
> Thank you both for your reply.
> 
> So how did you guys verify and test gromacs on your machines?
> Mark, from your response, test-set 4.0.4 is of no use?!  I have seen your
> other posts but the problem here is far more basic than that.
> 
> If it happens that 4.0.4 is usable to some degree, there's still the issue
> I brought up on my initial post - the script gives the grompp error.  Now
> looking at the code I can see where the problem is:  subroutine gmx_test()
> is called before double or single precision, or any of the options for
> that matter, are determined in the for loop following the gmx_test() call.
>  And so you end up with the "ERROR: Can not find grompp in your path...." 
> again, regardless of whether or not you have provided any options -
> besides without options it should respond with usage info which it
> doesn't.  However, in test-set 3.3.2, this subroutine does not exist and
> like I said in my previous post, it does run upto a certain point.  So,
> correct me if I am wrong, but gmx_test() is out of place in versions 3.3.3
> and 4.0.4.
> 
> Which takes me back to my question - how did you guys, or anyone else, do
> your testing.  I thought of modifying the script to make it run but it
> looks like it needs more than just moving that one line.  Also, I just
> can't believe that this type of modification is expected on stuff
> available for download on the site.
> 
> I mean, if this is really the case can one of the developers fix this and
> put up a usable script.  And, again, would doing so make the test set
> usable to some degree?
> 

No such modifications should be necessary.  The script should work "out of the 
box," so to speak.  Let me go back to my original question - do you have a 
single-precision Gromacs installation, or just double-precision?  I seem to 
recall you saying that you were testing by running luck_d only.  How are you 
calling the script?  Just "./gmxtest.pl all"?  Because doing so only tests 
single-precision.  If there is a problem testing double-precision, it certainly 
needs to be addressed.

Whether or not the tests are meaningful is a bit more complex, as you note. 
Failures aren't always failures, per the many threads in the archive on this 
topic :)  Since Mark did quite a bit of work some time ago, the test set has 
been much more robust, but as you'll note from our discussion, more features and 
combinations should be implemented.

-Justin

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================