[gmx-developers] XML files

Erik Lindahl lindahl at stanford.edu
Thu Mar 21 19:28:06 CET 2002

David van der Spoel wrote:

>On Thu, 21 Mar 2002, Erik Lindahl wrote:
>>Hi David,
>>I'm just starting to grasp the XML formats, so I'll do some reading up 
>>before I share any deeper thoughts about it.
>>One thing, though: Since we rewriting things anyway, shouldn't we get 
>>rid of all strange abbreviations in the input files? I.e., change things 
>>like 'nstenergy' to e.g. 'energy_file_output_frequency'.
>Yes and no. If you have to type it you want it to be short. The nested 
>format of XML makes it somewhat clearer already:
><gromacs title="A mixture of water and methanol">
>  <!-- inputrec holds variables for the dynamics runs -->
>  <inputrec algorithm="bd" nsteps="100" delta-t="0.002">
>    <output nstlog="10" nstenergy="1" nstxout="10" nstvout="10" />
>So you "immediately see" (tm) that output is a subset of inputrec and 
>inputrec a subset of gromacs. But looking at it: inputrec could be renamed 
>to parameters, and many others could be clarified too.

Actually, it would be nice to have units on all output and step stuff, 
so you could say that you want
100 ps of simulation with 2 fs timestep, and energy file output every 10 
ps. I don't think it affects the DTD, though, so we can do that later.

Is it possible to include help texts with a longer description of each item?

>This will be possible, my preference would be something like
>mdrun params.xml coords.xml topol.xml forcefield.xml
Ok. We don't really have to put any restrictions on the names - mdrun 
can just read 'everything' and complain if something is missing or 
present in more than one file. I guess we should use the last occurence 

>>I'll think of ways we can separate topologies from forcefields.
>The problem is merging topologies with force fields.

Agreed; but it would be darn nice to have! By the way, since you've been 
doing NMR stuff - I think it would be nice to be able to use atom names 
instead of numbers when describing bonds, etc. What's the most common 
way people would be adding NMR or X-ray restraints, and how do they 
select their atoms?

>>We also need to keep the old trajectory formats as optional output too 
>>until all programs like VMD, gOpenMol, etc, can read the XML formats. 
>>And, we should implement support for the new formats in babel... So, 
>>let's not remove any file types until we are sure everything else is 
>>working in the new formats!
>We'll keep trr and xtc obviously for space reasons, maybe we need to keep 
>gro for a while...
And some kind of file format that doesn't use RPC libraries..  Even if I 
rip the RPC stuff from linux we can't be sure it will work 'everywhere'. 
I don't have the slightest idea what will happen with 64-bit windows, 
for instance...



More information about the gromacs.org_gmx-developers mailing list