3 ms·

Why not go the SQLite way? Open source but don't accept external contributions. Literally just dump the code.

• cryptonector 2y ago
  > Open source but don't accept external contributions. That's not the key to the SQLite model. The key to the SQLite model is that their 100% code coverage testsuite is proprietary. You can't credibly fork SQLite3 w/o a similar testsuite because everyone knows that SQLite3 has that testsuite and so it's simply better unless your fork has one too. This works very well for SQLite because it is the single most widely used piece of software ever. And that is because it solves such an important and universal problem (a local DB RDBMS) with such convenience (embedded, server-less). The reason that SQLite does not accept contributions is not so much that they don't want to, but that contributors can't contribute changes to the proprietary testsuite, and writing those is harder than writing the contribution, therefore contributions impose a big tax on the SQLite dev team that they prefer not to pay. Very few other pieces of software have similar universal usage/applicability/convenience stories. Therefore it's not easy to apply the SQLite model to all the things. Sourcegraph could have a business source-available option. If all you want is to be able to be able to debug problems and/or make contributions and you're a paying customer, then why would that not be enough? SQLite is essentially source-available given that you can neither contribute to nor credibly fork it.