Dolt replica

[decalek]

@apavlo, Thank you for the amazing work of classifying the DB world!

What is your opinion on eventual replica of dbdb.io DB as dolthub.com/cmu-db/dbdb.io (in case that this stays aligned to your future plans for that collection of course).

Technically, after the initial myslqdump/load (as you probably remember, dolt server is mysql protocol compatible), we could think of some form of replication between the site DB instance and a dolthub "site" branch of the DB (e.g. via tarantool or debezium).

With fork-able replica, you then could receive eventual PRs both as schema improvements and data contribution, and users will gain the power to query the dataset with SQL (locally and on the hub).

BTW, The Dolt people are on they road to opensource the dolthub interface (the alpha is already available as free for use, closed source download), so at some point, we would be able to spawn a specialized hub for the dbdb.io (with eg. sourcegraph, glean etc. over the OSS DB implementations) plus the other datasets which you collect and provide to the community (e.g. the RSS feeds).

[apavlo]

I don't understand what you are asking?

[decalek]

@apavlo, Dolthub tries to be a "github" for relational data (supporting forking, pull-requests, etc.). It is based on the MySQL compatible versioned DB dolt.

Do you mind publishing a dolt repo with the collected dbdb.io data on dolthub?

[edude03]

I put together an example over here so that you can get an idea, but yeah an official version would be better of course.