Merge CIF coordinates into NeXus

[stuartcampbell]

Original reporter: prjemian

Merge the method used by CIF to define coordinates into the NeXus coordinate system.

[stuartcampbell]

Original reporter: zjttoefs

The manual already talks about e.g. the transformation_type attribute, but I cannot see that in any NXDL. Have I lost my grep skills or have we really not added that yet?

I was looking for the right place to put the dependency business.

[stuartcampbell]

Original reporter: prjemian

The transformation_type has not yet been added. That's part of the work for this ticket. Is there yet a clear description of what work should be done? Reference to this term is in manual/design.xml.

Perhaps we should pull the parts of the manual using transformation_type until its use is decided and built into NeXus.

[stuartcampbell]

Original reporter: prjemian

In [1062]:

#CommitTicketReference repository="" revision="1062"
refs #112, transformation_type is not yet built into NeXus, this part is getting ahead of ourselves

[stuartcampbell]

Original reporter: zjttoefs

I believe in Oak Ridge we decided to have transformation_type (and vector). What is open (and what I am currently working on) is recording the dependencies, i.e. order of transformations.

So rather than removing it from the manual I'd put it into the NXDL.

[stuartcampbell]

Original reporter: zjttoefs

Pete, I'd really like to get this going. But someone needs to make a call where this geometry needs to be defined. Should that be based in NXobject? That would be my guess.

[stuartcampbell]

Original reporter: prjemian

NXobject has no content. It is the base definition for the NeXus base classes.

The definitions should go into either nxdlTypes.xsd or nxdl.xsd. Please post a proposition or example in this ticket.

[stuartcampbell]

Original reporter: zjttoefs

An example would be:

         DATASET "tilt" {
            DATATYPE  H5T_IEEE_F32LE
            DATASPACE  SIMPLE { ( 1 ) / ( 1 ) }
            DATA {
            (0): 0.05
            }
            ATTRIBUTE "depends_on" {
               DATATYPE  H5T_STRING {
                     STRSIZE 47;
                     STRPAD H5T_STR_NULLPAD;
                     CSET H5T_CSET_ASCII;
                     CTYPE H5T_C_S1;
                  }
               DATASPACE  SCALAR
               DATA {
               (0): "/entry1/instrument/manipulator/azimuthal_angle"
               }
            }
            ATTRIBUTE "transformation_type" {
               DATATYPE  H5T_STRING {
                     STRSIZE 8;
                     STRPAD H5T_STR_NULLPAD;
                     CSET H5T_CSET_ASCII;
                     CTYPE H5T_C_S1;
                  }
               DATASPACE  SCALAR
               DATA {
               (0): "rotation"
               }
            }
            ATTRIBUTE "units" {
               DATATYPE  H5T_STRING {
                     STRSIZE 6;
                     STRPAD H5T_STR_NULLPAD;
                     CSET H5T_CSET_ASCII;
                     CTYPE H5T_C_S1;
                  }
               DATASPACE  SCALAR
               DATA {
               (0): "degree"
               }
            }
            ATTRIBUTE "vector" {
               DATATYPE  H5T_IEEE_F32LE
               DATASPACE  SIMPLE { ( 3 ) / ( 3 ) }
               DATA {
               (0): 1, 0, 0
               }
            }
         }

The depends_on is new, the rest should be in there already (but isn't) since it was agreed in Oak Ridge.

The idea is that a missing depends_on would signal that this transformation starts in the origin of the coordinate system. You could say that explicitly with depends_on=.

To fix e.g. an NXdetector in space a depends_on field would point to the last translation or rotation in the chain required to reach the location.

[stuartcampbell]

Original reporter: prjemian

A table in manual/source/design.rst has been commented out awaiting resolution of this ticket. Before closing the ticket, resolve the TODO item at or near line 606 in that file.

[stuartcampbell]

Original reporter: mkoennecke

In [changeset:1180]:

#CommitTicketReference repository="" revision="1180"
- Updated the manual to describe the CIF style transformation encoding 
  properly. refs #112

[stuartcampbell]

Original reporter: prjemian

In [changeset:1181]:

#CommitTicketReference repository="" revision="1181"
typos and formatting. refs #112

[stuartcampbell]

Original reporter: mkoennecke

In [changeset:1186]:

#CommitTicketReference repository="" revision="1186"
- Clarified (hpefully) an issue about the depends_on field. refs #112

[stuartcampbell]

Original reporter: zjttoefs

Pete, I need your input on how to represent the axis in NXDL. I am stuck there.

[stuartcampbell]

Original reporter: mkoennecke

In [changeset:1242]:

#CommitTicketReference repository="" revision="1242"
- Marked the polar coordinate system and NXgeometry as planned to be deprecated

Refs #112

[zjttoefs]

This is done, when #273 is done.