Until today, I kept the test results (diagnostic work) in one table :). Although I always understood that this was not normal. It is clear that sometimes you have to move away from some form of normalization, but not so much.

And now, after decomposition, I now offer this scheme to myself:

enter image description here

And good people, please make your comments.

I myself have a remark, but I can offer nothing else:

I don't like the fact that there are so many composite keys. Probably somehow it would be correct to use the type uniqueindetifier ? In the WorkParticips table WorkParticips I cannot use the uniqueindetifier , because I often need to transfer this ID to the participants themselves. Should the SubjectParticips and SubjectResults tables SubjectResults manipulated with the uniqueindetifier ? Or is there a serious problem in the proposed scheme?

  • I sit, I look at the scheme ... here is HOW you have to imagine the testing process in time and space to draw such a scheme for it? - Akina
  • @Akina, I'm sorry, but I didn't quite get the idea. Am I very good at drawing or very bad? :) - Adam
  • 2
    From my point of view, the scheme does not correlate with the process at all. Perhaps this should be classified as "bad." Try to forget that you have something (a table with data, an old analysis, etc.), and start everything from the very beginning, from scratch. IMHO you will also be easier. - Akina
  • Try to forget that you have something @Akina so I just do it. The fact is that the base is recognized to store only statistical information. No data about the procedure itself is in the database. - Adam
  • But this does not mean that the essence of the Work becomes an attribute of the essence of the Participant. - Akina

1 answer 1

Look at this option, without composite keys: enter image description here