@tillkruss Adding a fix for Relay connection, leads to an error in CI

    /**
     * @throws ConnectionException
     * @throws ClientException
     * @throws ServerException
     */
    public function connect()
    {
        foreach ($this->initCommands as $command) {
            try {
                $this->executeCommand($command);
            } catch (ServerException $ex) {
                if ($command->getId() !== 'CLIENT') {
                    throw $ex;
                }
            }
        }
    }

https://github.com/predis/predis/actions/runs/5678238422/job/15388057449?pr=1347

Trying to run Relay tests locally, but have a problems with relay installation over homebrew.

Error: An exception occurred within a child process: MachO::DylibUnknownError: No such dylib name: /usr/local/lib/libhiredis_ssl.dylib.1.1.0

This file actually exists and I have correct version of hiredis

Aah, maybe Relay is already setting CLIENT SET-NAME is causing the issue.

Trying to run Relay tests locally, but have a problems with relay installation over homebrew.
Error: An exception occurred within a child process: MachO::DylibUnknownError: No such dylib name: /usr/local/lib/libhiredis_ssl.dylib.1.1.0

Relay via Homebrew uses OpenSSL v3, can you try brew uninstall relay and reinstall it?

coverage: 77.719% (+0.04%) from 77.682% when pulling 15de538 on vladvildanov:vv-client-set-info-on-connect into df14e11 on predis:v2.x.

Thanks for your work on this library, it serves us well since years but this PR is problematic for us.

1. In this PR add 2 commands on initialization, that's a lot of IO. It should at least be an option IMO. We saw an important load increase in production due to this PR.
2. Why was it bundled in a patch version?

We saw an important load increase in production due to this PR.

Can you elaborate on this and provide numbers?

We didn't have "client" call before

And here you can see the substantial change in CPU in the cluster:

it took us a while to find the culprit because we didn't expect to have a change in behavior in a patch version

@vladvildanov @chayim Any thoughts?

@tillkruss I talked about this with other client-developers and they confirm that this should be an optional.

We can't fix it other way since it's a Redis API and we must call it twice each time. Let's merge #1370, but before we need to merge #1369 that fixes CI after Redis 7.2 update

@BafS I'm really sorry about that situation and I'm appreciate your feedback that helps to make Predis better. #1370 PR will make this optional and allows to have more control over it. @tillkruss I believe we can include this changes in our next patch version ASAP it will be merged?

Thank you for your feedback and thanks for helping fixing it.

Don't you think that this PR should be reverted for the next patch release (to revert the issue introduced) and the new feature (with the new set_client_info defaulting to false to not have BC) should be in a new minor version?

@BafS Totally works for me, but @tillkruss is the releases responsible person