Put ziggy.js in repo or .gitignore?

[plweil]

How are people dealing with ziggy and deployments? After integrating ziggy into my dev environment, I'm now preparing to deploy a test branch. Initially I've had the ziggy.js file in .gitignore, with the idea of generating a new ziggy.js on the server during deploy (e.g., by running the artisan task in capistrano). I can see now that this isn't going to work: webpack will spit out errors because it can't resolve the references to ziggy.js, which doesn't exist yet.

So the choices seem to be 1) put the generated, full-fledged ziggy.js file in the git repo; or 2) generate a blank ziggy.js file for the repo, and then run the artisan ziggy:generate command during deploy. Are there any real downsides to putting ziggy.js in git, even though it feels "wrong"? What are others doing?

[bakerkretzmar]

Can you run php artisan ziggy:generate before your Webpack build, so the file does exist?

[plweil]

Yeah, I think we just need to find where/when to run that artisan task. After composer, but before webpack/Mix.

[plweil]

We came up with a solution that runs the task as a composer post-install script. Which works fine for deployments. However...in the dev or local environment, you need to have a ziggy.js file in place before running npm install. This can cause an issue whenever you pull a branch to which someone has added/edited node modules. As far as I can see, you can either run the task manually before running npm install (which is a little inconvenient), or possibly keep an empty, "placeholder" ziggy.js file in the repo. Either option feels a little hackish. I wonder what others do; I can't be the first person to run into this.

[bakerkretzmar]

You shouldn't need a ziggy.js file to run npm install, do you get some kind of error if you try to do that?

[plweil]

Sorry, I should have written that more carefully. We have npm run production as a postinstall script. So it's really a webpack error.

[bakerkretzmar]

Gotcha. Yeah I think you should just commit the whole ziggy.js file to the repo, way easier that way and I don't think it's 'wrong'.