Disable example_status_persistence_file_path

[technicalpickles]

Mentioned this on #226 (comment)

I did find another benefit to doing this. If you have config.example_status_persistence_file_path set, you end up loading and dumping the status for all specs each time.

We ended up disabling this during CI. I could see it being a knapsack pro default if there is a good place to set it.

Documentation about this feature:

The file path to use for persisting example statuses. Necessary for the --only-failures and --next-failure CLI options.

I came across this while investigating #200 , and I saw a lot of memory allocations from the code that generates and publishes this file. Since knapsack wouldn't be using --on-failures or --next-failure, it should be safe to disable these.

[ArturT]

What do you mean by disabling example_status_persistence_file_path? Is it about getting rid of some internal RSpec code that is responsible for handling that?

If a user does not set example_status_persistence_file_path in their spec_helper.rb, does it mean RSpec is still consuming a lot of memory?

We have some customers using --only-failures option to retry flaky tests. This is documented here.
This means they need to set example_status_persistence_file_path.

[technicalpickles]

I mean not setting config.example_status_persistence_file_path, or setting it to falsey

If a user does not set example_status_persistence_file_path in their spec_helper.rb, does it mean RSpec is still consuming a lot of memory?

The code path that allocates a lot of memory doesn't get used. Specifically, it's the persister and loader in https://github.com/rspec/rspec-core/blob/1c662c35f91f33946fdfbed431878f4798f06f70/lib/rspec/core/example_status_persister.rb .

We have some customers using --only-failures option to retry flaky tests. This is documented here.
This means they need to set example_status_persistence_file_path.

Ah, in that case, this wouldn't work for them. Given that, I don't think it's worth adding complexity code-wise or documentation-wise to specify it. I think we can leave it as is then, and let users identify and disable themselves. Or #226 lands and it's mitigated because it's only persisted/loaded once

[ArturT]

Ok. So no action for now.

By default config.example_status_persistence_file_path is not used by the majority of users so they have no memory consumption issue.

Some customers use config.example_status_persistence_file_path and they use RSpec in Queue Mode and this can cause increased memory consumption. This is the issue you discovered.

When we merge PR #226 then the issue won't exist anymore for the customers. So let's focus on the PR.

I'm closing this one for now.