adding phase_id and/or diffractogram_id to categories as keys

[rowlesmr]

as mentioned in #79

Adding phase_id and/or diffractogram_id to categories as keys to allow multiple appearances of the various categories in difference blocks in the same data collection.

[rowlesmr]

• PD_CALIBRATION: I think needs a diffractogram id. It talks about calibrating _pd_meas.detector_id or _pd_meas.position values to Q, energy, angle, etc.
• PD_CHAR is a funny one; even though it is about the specimen, I think I to want a link to diffractogram_id.
• PD_PREP: I think this needs to link somehow with PD_SPEC. PD_PREP details the preparation of the material from which the specimen is drawn, whereas PD_SPEC details the actual thing you put in the beam.
• PD_PEAK: I don't know enough about it to know if its to be treated as a phase or a diffractogram.
• PD_PEAK_OVERALL: I think a diffractogram id.
• PD_PROC_LS is difficult. In the trivial case, it depends on one of each id, in the normal case, one diff_id and many phase_ids, and in the pathological case, many of both ids.

[rowlesmr]

The following already have a _pd*.detector_id as a category key (although I still need to work out how that ends up pointing to a diffractogram_id)

• PD_CALIB
• PD_CALIB_OFFSET
• PD_CALIB_STD
• PD_INSTR_DETECTOR

Maybe I can also give _*.diffractogram/phase_id to these categories, and the detector_id may be able to be found through them (just not added as category keys).

[rowlesmr]

When adding keys, I'm thinking

If I put this Set/Loop information in a data block all by itself, how do I get the information back into the diffractogram and/or phase of interest.

Is that the right frame of mind?

[jamesrhester]

When adding keys, I'm thinking

If I put this Set/Loop information in a data block all by itself, how do I get the information back into the diffractogram and/or phase of interest.

Is that the right frame of mind?

Yes, sounds OK. You can also think "would these values necessarily change if the phase/diffractogram/detector/X was different, and is that already captured via the key data names I already have".

[rowlesmr]

These are a little different to the other categories:

• PD_CHAR: information about the 'material'.
• PD_PREP: information about the 'sample'.
• PD_SPEC: information about the 'specimen'.

They're all Set categories, and form a hierachy (in the above order), with only PD_SPEC actually having any formal link to a diffractogram. I don't think any of them have a link to a specific phase.

In a formal sense, I think that

• PD_CHAR is keyed on _pd_char.id;
• PD_PREP is keyed on _pd_prep.char_id, and _pd_prep.id; and
• PD_SPEC is keyed on _pd_spec.char_id, _pd_spec.prep_id and _pd_spec.diffractogram_id (and possibly even _pd_spec.id if you reuse the same specimen to make different diffractograms).

That seems rather complicated.

In a pragmatic sense, I think that

• PD_CHAR is keyed on _pd_char.diffractogram_id;
• PD_PREP is keyed on _pd_prep.diffractogram_id; and
• PD_SPEC is keyed on _pd_spec.diffractogram_id.

is possibly a way to go.

How could you then deal with an in situ experiment with many diffractograms with the same specimen (and hence, the same sample and material)?

[rowlesmr]

I think all the easy ones are done. The remainder, I think, need some thought.

eg PD_PEAKS: to whom do they belong? phase? diffractogram? both? what about _pd_background.peak_id?

[jamesrhester]

I think (after those two comments of mine are resolved) we can merge these changes, and then tackle the other categories you've raised as separate small pull requests.