Re: Derived logical volumes in VGM

 Hi Brett,

 Currently, user custom solids are not supported;  I want to provide hooks in VGM so that user can extend  it himself with his customized classes, but it will take  some time.
 Tolerating unsupported shapes can be done quickly;  I will try to make a new version enabling this during  next week. This approach is already used in exporting  geometries in XML, as at the time of development of VGM,  GDML has a lot of gaps. Here, unknown shapes are  placed as a small boxes with the correct transformation  and they are preceded with a warning so you can easily  trace them in the file. This seems to make sense also  for your use case.

 Best regards,

 Ivana

On Fri, 13 Oct 2006, Brett Viren wrote:

> Hello Ivana,
>
> Ivana Hrivnacova <ihrivnac_at_mail.cern.ch> writes:
> > Or you can change the line 239 in
> > vgm/packages/Geant4GM/source/volumes/Factory.cxx:
> > << solid->GetName() << "\")" << std::endl;
> > to:
> > << solid->GetName() << "\")" << " type: " << solid->GetEntityType() << std::endl;
>
> This reports that it is a "dywTorusStack" which is a custom class
> of ours that derives from G4CSGSolid.
>
> >> or at least to have VGM ignore tany unknown solids?
> >
> > There is a question what user could then do with such
> > incomplete geometry. But I am already considering this
> > possibility to make easier to get rid off problems when
> > a user hits some limitation as in your case.
>
> In this particular case we can make due with incomplete geometry.
> These custom classes are for our detailed PMT model. We only need
> this level of detail in our MC. For reconstruction codes we don't
> need to know about the details of the PMTs just where they are placed.
> My thought was to store just PMT locations in VGM and leave their
> detailed construction hard coded in the MC codes.
>
> Does it make sense to try this approach?
>
> > For dumping geometry in XML I suggest to use rather
> > AGDD exporter than GDML, which is more complete.
>
> Ah, will do.
>
> Thank you for your help,
> -Brett.
>

-- 
==============================================================
e-mail:  Ivana.Hrivnacova_at_cern.ch
address: Institut Physique Nucleaire, 91406 Orsay, France
phone:   (33) 01 69 15 65 94
==============================================================