Understanding the difference between generated and _generated

[PaulRBerg]

I've noticed that we're using both generated and _generated. Why?

And a question about this path:

indexers/.gitignore

Line 7 in 07b3166

!apps/*-envio/src/generated

Shouldn't this be _generated instead of generated?

If not, why do we keep src/_generated in version control?

cc @razgraf @gavriliumircea

[razgraf]

src/_generated has to be kept in because

• it brings the contract addresses from the shared package into the indexer package
• its values are used at both build time and run time (check imports from those location in the codebase)

When we add a new address, before deploying, we need to run a pre-deployment step (see the setup script) which includes this generation step. We need to do this because envio and the-graph use different package managers (pnpm and yarn) which don't play nice with shared workspaces and direct external imports (I tried). Hence the need for a copy-paste step.

You are correct that the rule should say _generated (we want to force it to be included in the version control). We can't use generated as a name because it conflicts with the actual envio generated build files (which are housed in *-envio/generated). Figured that out later on so most likely forgot to update the rule when I renamed that folder with the underscore prefix.

[PaulRBerg]

Got it, thanks for the explanation.

In this case, we will no longer need this directory (src/_generated) and any associated functionality because the @sablier/deployments package provides all the data.

However, if I'm wrong about that and we will continue to need it for another reason, I suggest renaming this directory to something else — the current name makes it seem like it is analogous to the generated stuff in the subgraphs, but this is not so. The data are not analogous at all, hence why _generated should NOT be git ignored, while generated should be git ignored.