[Nexus] item discussed now

[Eugen Wintersberger]

Hi Tobias

On Wed, 2015-02-18 at 13:15 +0000, Tobias Richter wrote: 
> Hi Eugen,
> 
> The centre is the lowest common denominator. It mostly works for the NXsample and possibly for NXslit, I would say.
> For all other cases ideally we need to define a prominent reference point. Like first pixel for NXdetetcor. That’s best done case by case.
> In an abstract way I cannot guarantee that we’ll always find a suitable generic location independent of the specific model or construction. But we can try.
> 

I entirely agree with you. As I have a talk at the PETRA III Extension
Meeting next week I can use this opportunity to discuss this issue with
our beamline scientists. Maybe we ca agree upon something ;)

regards
  Eugen


> Regards,
> 
> Tobias
> 
> 
> > On 18 Feb 2015, at 14:03, Eugen Wintersberger <eugen.wintersberger at desy.de> wrote:
> > 
> > Hi folks
> >  though that I did not attend todays telecon - I would like to make a
> > remark here as I am currently working with transformations quite a lot
> > while designing the NeXus layout for several of our beamlines at PETRA
> > III.
> > 
> > On Wed, 2015-02-18 at 07:36 -0500, Herbert J. Bernstein wrote: 
> >> As per our discussion on the hangout today, please note the following
> >> clarification:
> >> 
> >> When specifying the position of a component with a depends_on field,
> >> the specific point in the component being positioned will either be
> >> the center of the first pixel for a detector 
> > 
> > This is quite fine. One can mange this as every 2D detector has
> > something like a first pixel. 
> > 
> >> or the center of the
> >> component for components not having an equivalent to a "first pixel".
> > 
> > This is a bit more complex. What would be the "center" of a component. 
> > The entire situation remains me a bit on 3D modelling software which
> > follows a composite geometry approach. All of their 3D primitives have a
> > well defined center. However, one has to consult the manual of the
> > software package to learn where the center for a particular primitive is
> > (different software packages can make different choices where to place
> > the center for a particular primitive). At least all software packages I
> > kow use this "center" as the origin for the primitive local coordinate
> > frame. In my opinion this is what we are missing - a center for each
> > "component" onto which we can bind the origin of the local coordinate
> > frame.  
> > 
> > The question that arises for me is: is it possible to define such a
> > center for our base classes (for those who can be used along with
> > transformations) in a generic manner (in a way independent of the
> > particular mechanical construction)?
> > 
> > BTW. This discussion does not render transformations useless. From my
> > experience at PETRA III they have already proven quite useful when
> > describing the dependencies of axes. However, I guess this discussion
> > can greatly improve the usability of transformations.  
> > 
> > regards
> >  Eugen
> > 
> >> 
> >> On Wed, Feb 18, 2015 at 7:29 AM, Pete Jemian <prjemian at gmail.com> wrote:
> >>> 
> >>> use this URL for the transformations
> >>> http://download.nexusformat.org/doc/html/design.html#coordinate-transformations
> >>> 
> >>> On 2/18/2015 6:10 AM, Pete Jemian wrote:
> >>>> 
> >>>> https://github.com/nexusformat/definitions/blob/depends_on/nxdl.xsd
> >>>> 
> >>>> 
> >>>> On 2/18/2015 6:05 AM, Pete Jemian wrote:
> >>>>> 
> >>>>> 
> >>>>> 
> >>>>> https://github.com/nexusformat/definitions/issues/273#issuecomment-74158201
> >>>>> 
> >>>>> 
> >>>> 
> >>> _______________________________________________
> >>> NeXus mailing list
> >>> NeXus at nexusformat.org
> >>> http://lists.nexusformat.org/mailman/listinfo/nexus
> >>> 
> >> _______________________________________________
> >> NeXus mailing list
> >> NeXus at nexusformat.org
> >> http://lists.nexusformat.org/mailman/listinfo/nexus
> > 
> > 
> > _______________________________________________
> > NeXus mailing list
> > NeXus at nexusformat.org
> > http://lists.nexusformat.org/mailman/listinfo/nexus
> 
> 
> _______________________________________________
> NeXus mailing list
> NeXus at nexusformat.org
> http://lists.nexusformat.org/mailman/listinfo/nexus