[Dune-devel] Updates on the PSurface project

[Markus Blatt]

HI xiaoxue,


On Sat, Aug 03, 2013 at 01:02:22PM -0400, Xiaoxue Gong wrote:
> Last week I modified the vtu and hdf5 io code according to Markus'  guide
> to make it  looks better:
> 1. Changed the method and class names according to Camel-case  to adhere to
> dune guidelines.
> 2. Remove all the indent and trailing white space in the code.
> 3. Put all the functions that deals with io into one class(vtkPWriter).
> 4.Extend the template parameters of vtkPWriter. It could only deal with 2
> dimensional psurface and float. Not it support 1 dimensional psurface, and
> the float.

Are you sure that you pushed all of your changes? I do not see a way
of using a PSurface with dimension 1. Could you explain how to
read/write such a surface with the current code?

> 5. Use command line parameters to get the input filename.
> The modified code have been pushed to git://
> gitorious.org/gcos/psurface-convert.git
> 
> Issues that I have not solved includes:
> + Moving headers
> "Some your headers seem to be usable by other programs. E.g. one might want
> to read xmf files in one's own applications. This headers should be under
> dune and be installed, the cpp file should become part of a library. You
> can see how this is done in dune-common/dune/common. "
> +The include path issue
>   What is up with the weired AC_ARG_WITH clauses in configure.ac? I do not
>    think that they are needed and doubt that they serve their intended
>    purpose.
> +Please try to document your code along (or even before) writing it
>    using the doxygen syntax. Most developers (me, too) are not
>    disciplined enough to document their code after writing it. In the
>    rare occasion the we are and that there is time left, we are often
>    having a hard time understanding it ourselves (which is a shame and
>    waste of time). I would recommend documenting it properly _now_.
> 
> I will address these issues in the next week  and then move to task #2:
> "start with input from a gmsh file and construct a PSurface object from a
> given domain surface."

Does that mean that reading a psurface from HDF5 is already
implemented? If not then please first finish 3.

> In order to to that I need a gmsh file to start with. Maybe I should first
> write functions that can write PSurface object into gmsh file.  The file
> should contain all information I need to build psurface.

No you should not. I do not think that there is any need to be able to
write a PSurface object to GMsh format. You can use gmsh itself to
create a parameterized surface.

Regards,
Markus

-- 
Do you need more support with DUNE or HPC in general? 

Dr. Markus Blatt - HPC-Simulation-Software & Services http://www.dr-blatt.de
Hans-Bunte-Str. 8-10, 69123 Heidelberg, Germany
Tel.: +49 (0) 160 97590858  Fax: +49 (0)322 1108991658