Open NMR: what metadata do we want?

One of the reasons that CrystalEye works is that the metadata contributed by the authors (and required by the publishers, through IUCr) is superb. Is there general agreement about what metadata should be captured for NMR spectra or shifts? The JCAMP files potentially contain a lot but this depends on their availability – we know of few publishers who accept, let alone publish JCAMP files. On the assumption that this post reaches some publishers who wish to promote good scientific practice in reporting spectroscopy, what do we want.
NMRShiftDB mirrors the summary material  published in the body of chemical papers (and sometimes supplemental). A typical recent entry shows:

Bookmarks You could bookmark structures if you were logged in!

Details Spectral Data Additional Data Molecule (20139616) nmrshiftdb.cubic.uni-koeln.de_Kolshorn_2007-10-23_02:02:58_0723 Chemical name(s) (5-bromo-2-furyl)methanol Molecular weight 176.996 Number of all rings, size of smallest set of smallest rings 1, 1 CAS-Number   Molecule keywords   Type 13C Assignment Method 1D shift positions Solvent Chloroform-D1 (CDCl3) Additional comments MZ2N-107 Spectrum categories ocmainz inhouse database

The following may be available elsewhere but (for 13C) I would like to see additionally:

• organization/person depositing data (i.e. not just in filename)
• date of deposition
• whether solvent is spiked with reference material (e.g. TMS).
• method of assignment of peaks (ideally this should be per peak as well as per experiment)
• known hydrogen counts (this might be from in-house experiment or report in literature).
• comments

A lot of this can be gleaned from publications. Here's a typical rubric:

The ¹H NMR (300 MHz) or ¹³C NMR (75 MHz) and DEPT spectra were recorded in CDCl₃ using Brucker 300 MHz or JEOL 60 MHz spectrometers with TMS (0 ppm) as the internal standard.

PMR: but that's all you get. At least we know that they used real TMS.