Xavier Noria

Generally speaking, I believe parent applications are responsible for that kind of performance tweaks. Parent applications are the ones that know what is more suitable in a given runtime. In...

Agree with all the points made by Jean. It's one of those situations in which a trade-off has to be made and move on. If the default is not optimal...

If we go lazy by default, I'd generate the `CI` line at the bottom (off-the-top of my head, feel free to edit): ```ruby # Client code may eager load the...

I guess the problem is that the rule says "Google" instead of "Googlebot". The patch does unnecessary additional edits, I think, would you mind doing that small edit only?

> My concern is that this introduces a new pattern for emitting events in the framework, where the other PR follows the established pattern of (emit notification -> event subscribers...

Of course, open also to what @gmcgibbon, @adrianna-chang-shopify, and @rafaelfranca think, they have the vision for events.

@adrianna-chang-shopify @Edouard-chin let me give you my point of view. Active Support Notifications were historically introduced to provide a framework for APMs. So, the method is `instrument`, not `publish`, or...

Hey, just a heads up that I have this pending to reply.

> But as a user, how would I know that beforehand? You know about the notification name and type because the docs tell you about it (that is how you...

Another option could be to standardize on structured events. In this alternative, the framework sends structured events for everything. Some of them may also be instrumented (meaning, there is an...