Predicates for is-SOMA and is-SOMA-collection

[johnkerl]

Context: #163

Example:

import tiledbsc

for path in [
  '/Users/johnkerl/mini-corpus',
  '/Users/johnkerl/mini-corpus/atlas',
  '/Users/johnkerl/mini-corpus/atlas/acute-covid19-cohort',
  '/Users/johnkerl/mini-corpus/atlas/acute-covid19-cohort/obs',
]:
    ans1 = tiledbsc.util.is_soma_collection(path)
    ans2 = tiledbsc.util.is_soma(path)
    print(ans1, ans2, path)

False False /Users/johnkerl/mini-corpus
True False /Users/johnkerl/mini-corpus/atlas
False True /Users/johnkerl/mini-corpus/atlas/acute-covid19-cohort
False False /Users/johnkerl/mini-corpus/atlas/acute-covid19-cohort/obs

[bkmartinjr]

Slightly related to this, while we are adding metadata do you anticipate also needing an encoding version number? Put another way, is there benefit to encoding the file format version so that the underlying TileDB object organization can evolve separately from the API?

[Shelnutt2]

Slightly related to this, while we are adding metadata do you anticipate also needing an encoding version number? Put another way, is there benefit to encoding the file format version so that the underlying TileDB object organization can evolve separately from the API?

Adding version numbers is usually the best thing to do. By the time you want to use the version number its usually too late and things are in the wild. Starting with one and not using it doesn't really cost much of anything.

[bkmartinjr]

fully agree! To my mind, this one is a "Tiledb instantiated (file format) for SOMA" version, and should be shared by all implementations on top of TileDB (eg, all languages/environments)

[johnkerl]

Thanks @bkmartinjr and @Shelnutt2 ! I'll merge this PR to avoid rebases etc but will put up another for versioning 🙏