Identification #160
Description
As of now, we informally use the value of repo as the quasi unique identifier for a project.
While repo itself is somewhat misleading, as it is often not what IT people think of repository (e.g. a SCM repo), but e.g. a Wiki page or even a ZIP file hosted on google-drive, the issue goes deeper:
It is totally possible and with hardware projects even very common, to have multiple repos, or say, multiple hosting sites for a single project. How would we decide, which one of them is the ID?
Would we completely ignore the existence of the others in OKH?
Also, while we could be lenient with what we accept as values in the (TOML) manifest, we should probably be stricter, or rather more precise on the RDF side of things. For example, which URL of a GitHub repo should be stored: Web-UI URL, HTTP(S) clone URL, SSH clone URL, ...
Yet an other issue we did not explicitly express or look at so far, is the difference between a project hosting and a project reference. For example: All OSHWA listed projects are really just references to projects hosted elsewhere, e.g. on GitHub or Appropedia. If we scrape both the OSHWA reference and the hosting platform, how do we handle the results? should we just ignore the OSHWA reference then, or rather the hosting platform one? what if OSHWA has additional info? which one is the primary one? how do we handle multiple references on different platforms to the same project? How do we cleanly gather and organize all that data, preferably in an automated way? - because nobody will correctly handle, and update all this info by hand anyway.
And if we handle all this in all its complexity and supporting all the possibilities out there, will people still be able to make sense of our data?
Will they understand how they will have to make queries?
Will it complicate querying the data a lot?
Related: