kuzdogan
commented
1 month ago Looks good. However, I don't get why we don't use RepoV2 for reading. Could you clarify that?
Now that you ask I'm thinking we could use it, maybe @marcocastignoli remembers and can enlighten us when he's back. But I guess it was about us saving file names in keccak hashes?? I mean we can still work around that. Maybe we were lazy and thought repoV1 is ok for now.
Turning off writing to DBs to be able to do a release because the DB schema on prod. is not updated and the release would fail to write contracts. The repo writing logic must be unaffected by the latest changes. We are doing an earlier release to add new chains. The contracts after this change will get written to repoV1/2 and we will add them later during the migration.
And using repoV1 as the read source since writing to the DB is turned off.
Similarly the VerA DB schemas are still not ready. It is ok to have writing to VerA turned off on staging but it would be nice if we could get the VerA schema ready before the prod. migration so that we write everything simulatanously to Sourcify and VerA and don't have to move rows from Sourcify DB to VerA