Nothing sign posts remote builds as the solution to large build errors.

[frostebite]

Something should more clearly state in the GitHub docs that larger repository sizes will require either a self-hosted runner or remote builder.

Perhaps we should catch the error and output a link to the docs also.

[webbertakken]

I do not understand the description right now @frostebite.

[frostebite]

Sorry about that. Updated the description. Let me know if that's any clearer.

[webbertakken]

Gotcha!

Are we talking about the 5GB cache limitation in GitHub for project/Library here or about disk space usage of 18GB(~60GB)?

[frostebite]

This is referring to the fact that currently a user exploring the documentation about game.ci/github-builds will not be warned ahead of trying to run their project that maximum disc space is available. This will potentially lead to a new user first configuring the default game.ci/unity-builder, running out of space and then reactively discovering game.ci/remote-builder.

We should try and enable the user to understand this consideration early if possible and help them to avoid wasting time.

So this is referring to disc space usage. Not cache limitations.

[frostebite]

Is there somewhere on the getting started page we could sign post users to "cloud runner" / remote builder that you'd agree is fitting for that page?

@webbertakken

"If you have a project larger than x GB, you may want to take a look at Cloud Runner." Somewhere for example.

[webbertakken]

Good idea.

My suggestion would be to make a short section following the Overall steps section where you'd quickly choose one of the two just for the mental model.

In the cloud builder section we could explain when exactly to choose it, in more detail.

As for the criteria. I believe the optimum workflow differs on a case by case basis. My suggestion would explain the bottleneck, and giving people the choice of which runner to choose.

Perhaps something like this?

GitHub's cache action allows for a maximum per-cache size of 5GB, after which the cache may become less effective. If you face any bottlenecks, we would advise to switch to cloud runner.

[frostebite]

Great suggestion, will seek to include this in the documentation update.