sandbox.gsd creation about gsd-tools

[kurtseifried]

TEST data: how do we ensure the test data cannot be used in production, e.g. "example.org" like "example-data-NAME"? Use random strings for package names (that will be impossible to eyeball and match up, ugh)? Stick emojis into the data? Use 1900-1920 as the year?

[raphaelahrens]

What about a special branch (orphan branch) or better a separate repo?

[kurtseifried]

we can't use "example" because a lot of stuff uses that name:

https://pypi.org/search/?q=example
https://www.npmjs.com/search?q=example

1. I think explicit meta data (e.g. going back to the "what is this, a vulnerability? incident? test-data")
2. Titles/descriptions can't really be used, e.g. if we use "this is a test data entry" then what happens if that string ends up legitimately in a real entry?
3. We can't just use dates in the distant past, we want test data that includes current dates, and the test data will be updated (e.g. y2038 related problems? only 16 years away...)

I think having said this we add a giant root level tag labelling this as TEST DATA that cannot and will not ever occur in production data. Then we don't have to be all clever and try to come up with magic strings/dates/etc.

[kurtseifried]

As far as using a special branch/etc we need the actual data to be clearly test data because people will accidentally pull it in otherwise.