I think there was a discussion about code (instance content code, like sql and source setup not the Dremio code itself) being stored in git. So that you could git grep it etc.
Has there been any thoughts on this?
Hey @cottreld, makes sense – we’re considering a few approaches here. If you have specific thoughts or examples around how you’d like this to work, would be great to hear. Otherwise, we’ll reach out once we have more concrete design/timing.
I would generally organize everything in (git) repos. Probably there is an Dremio API that I am not yet aware of (or in development) that will allow the users to follow the convention they want. i.e. users write their own API wrappers and set the local conventions. For example, this could be SQL queries in json files or in the code itself.
For those that want to write queries in the UI, it could simply be up to them to push (suggest simply copy/paste) the query state into the repo in the locally correct way.
I would even go as far as having the UI drop things regularly to enforce that there is no state accrual not captured in the config or code repo. That is one of my greatest fears. The state must be in the git repos and “data”. This is not so much for control, but for discoverability and understanding through ‘git grep’, ‘git blame’, ‘git log’ etc CLI
1 Like