[Q-e-developers] XML update and constrained DFT code submission

[Paolo Giannozzi]

Hi

thank you for your patch for the old XML format, and sorry for the
inconvenience: I try to keep the old file format alive for the time
being, but I don't check it any longer.

I think I already fixed yesterday evening (CET time) the other problem
you report. It is clearly related to a minor but far-reaching change I
did: all variables containing directory names are now assumed to have
a '/' at the end. I couldn't stand any longer the way it was done
before: half of such variables ending with '/', half not, // '/' //
everywhere, and the same filenames generated 1001 times in 1002
different ways.

About new vs old file format: as long as the CDFT code uses QE
routines to read and write files, the file format should be
transparent, modulo occasional glitches. It will take some time before
the new format is completely stabilized, but no more than a few weeks.
To be done:
- write the charge density in G space and in a single file for charge
and spin polarization
- consider adding to binary files other quantities needed to use them
without other data (e.g., list of G-vectors for \rho(G), list of k+G
vectors for \psi(k+G))
- delete other binary files (gvect, gkvect)
- remove iotk binary from binary files, provide equivalent hdf5 files
- replace iotk with FoX
- ...

Paolo

On Thu, May 11, 2017 at 11:37 PM, Matthew Goldey
<matthew.goldey at gmail.com> wrote:
> Hi everybody,
>
> We're almost ready to submit our constrained DFT code to QE since our paper
> was finally accepted, but there's one major problem - the new XML format has
> a bug that prevents me from fully integrating with the latest IO routines.
> Using -D__OLDXML, everything works fine. The bug is illustrated for a normal
> DFT run in the attached test.tar.gz and was present in commit
> 9c55c8be77442d00b4aaf01b43311200fc15f826 (from yesterday). This test case
> works with -D__OLDXML but not with the new XML routines. There's an issue
> reading in XML files when nspin=2 in some (but not all) postprocessing
> routines.
>
> Should we submit the code using the old XML routines or wait for the new XML
> routines to become stable?
>
> As a follow-up, I should mention that the latest git commit needs a minor
> patch (attached as oldxml.patch) to compile with -D__OLDXML.
>
> Regards,
>
> Matthew Goldey
>
> _______________________________________________
> Q-e-developers mailing list
> Q-e-developers at qe-forge.org
> http://qe-forge.org/mailman/listinfo/q-e-developers
>



-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222