[gmx-developers] Fwd: renaming Gromacs binaries

[chris.neale at utoronto.ca]

I haven't experienced any naming problems (is there really another  
tool called grompp?), but I suppose that if people are, then that is a  
strong argument in favour of changing the names.

The converse argument: I compile the gromacs that is used by a number  
of people. Since I am aware of this name changing, I will create my  
own links so that users can access files as they used to. However, not  
all people who compile gromacs read this list and could end up getting  
annoyed with the "why did my scripts stop working?" emails that they  
are bound to get.

Here's an example of how the non-existence of mdrun could spell  
disaster (and did for me recently):

STARTTIME=$(date +%s)
for((;;)); do
   THISTIME=$(date +%s)
   stoprunning=$(echo "(${THISTIME} - ${STARTTIME})/3600 > 40"|bc -l)
   if((stoprunning)); then
     break
   fi
   cp md1.cpt backup_md1.cpt.${RANDOM}

   mdrun -maxh 6 ...
   #analyze the .xtc
   #trjconv -dt 50 the .xtc
   #rm the original .xtc
done

This is taken from a real script of mine that I really used to  
accidentally crash a GPFS filesystem a few days ago. The cp md1.cpt  
backup_md1.cpt.${RANDOM} line is intended to avoid bugzilla 390.

If the mdrun command doesn't work, then what you get is not nice  
checkpointing and analysis on the fly, but instead you copy the .cpt  
file to a new name every second or so for 40 hours. Submit a few of  
these at once and the load is too much for the filesystem to handle.

So changing scripts should be easy, but the consequences are not  
always obvious -- especially when many users use scripts written by  
somebody else int heir group.

Note: I got an email failure trying to send this a minute ago, so I  
appologize if this is the second such email.

Chris.

Quoting David van der Spoel <spoel at xray.bmc.uu.se>:

> On 2010-06-09 01.50, hess at sbc.su.se wrote:
>> Hi,
>>
>> I agree with Erik on all points.
>>
>> I also agree on the preference for g_...
>> Not only because it is shorter, but also because this causes
>> the least changes.
>
>
> And what about those voices (Chris Neale, Peter Kasson) that are
> vehemently against changing names?
>
> There is one serious problem with changing names, and changing all your
> scripts, and that is that it becomes difficult to compare to older
> versions of gromacs (unless you also change names of binaries in old
> gromacs installations).
>
> I'm happy to provide a simple perl script for patching scripts if that
> is of any comfort - personally I think that is the way to go.
>
> Chris, Peter, more comments?
>>
>> Berk
>>
>>
>>> Hi,
>>>
>>> I don't really have any super-strong opinion, since we usually use AFS and
>>> have separate modules for versions, etc., However:
>>>
>>> 1. We can have arbitrary prefix/suffix modifiers, but let's not do
>>> anything more complex (that will be a pain to implement both in autoconf
>>> and cmake)
>>>
>>> 2. I'm a bit hesitant to do any large changes at this point, since we're
>>> gearing up for some major rearrangements as soon as we've pushed out the
>>> 4.5 release. We might make grompp optional, use a new tool for fixing up
>>> structures, merge/change analysis tools, and then it also seems natural to
>>> decide on a new well-designed naming scheme at that point?
>>>
>>> 3. Long-term, I think we need to decide: Either we have a _small_ number
>>> of programs that get installed e.g. in /usr/local/bin, or a larger number
>>> in a separate directory.
>>>
>>> 4. If I had to choose, my preference would be towards g_XXX right now...
>>>
>>> Cheers,
>>>
>>> Erik
>>>
>>> On Jun 8, 2010, at 7:35 PM, David van der Spoel wrote:
>>>
>>>>
>>>>
>>>> -------- Original Message --------
>>>> Subject: renaming Gromacs binaries
>>>> Date: Tue, 8 Jun 2010 08:00:57 -0700
>>>> From: Peter Kasson<kasson at gmail.com>
>>>> Reply-To: kasson at stanford.edu
>>>> To: David van der Spoel<spoel at xray.bmc.uu.se>
>>>>
>>>> Hi David,
>>>>  I'd put in a strong vote for having an easy way to maintain the
>>>> current naming scheme if desired (although it might not be the
>>>> default).  There are a lot of scripts out there that will get broken,
>>>> etc...
>>>>
>>>>
>>>> Thanks,
>>>> --Peter
>>>> --
>>>> gmx-developers mailing list
>>>> gmx-developers at gromacs.org
>>>> http://lists.gromacs.org/mailman/listinfo/gmx-developers
>>>> Please don't post (un)subscribe requests to the list. Use the www
>>>> interface or send it to gmx-developers-request at gromacs.org.
>>>>
>>>
>>> ----------------------------------------------------------
>>> Erik Lindahl<lindahl at cbr.su.se>
>>> Professor, Computational Structural Biology
>>> Center for Biomembrane Research&  Swedish e-Science Research Center
>>> Department of Biochemistry&  Biophysics, Stockholm University
>>> Tel: +468164675 Cell: +46703844534
>>>
>>> --
>>> gmx-developers mailing list
>>> gmx-developers at gromacs.org
>>> http://lists.gromacs.org/mailman/listinfo/gmx-developers
>>> Please don't post (un)subscribe requests to the list. Use the
>>> www interface or send it to gmx-developers-request at gromacs.org.
>>>
>>
>
>
> -- 
> David van der Spoel, Ph.D., Professor of Biology
> Dept. of Cell & Molec. Biol., Uppsala University.
> Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205.
> spoel at xray.bmc.uu.se    http://folding.bmc.uu.se
> -- 
> gmx-developers mailing list
> gmx-developers at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-developers
> Please don't post (un)subscribe requests to the list. Use the www
> interface or send it to gmx-developers-request at gromacs.org.