Component that has error is not being rendered

[rockwellll]

Hi there, hoping you're having a good start of the week!

I have been bothered by the issue for a couple of time now, since upgrading to Ruby 3.3.0 and Rails 7.1.3. If a component has some sort of error in it, i.e logical, or undefined parameters, instead of raising an error so I can see where things are going wrong. ViewComponent simply does not render the component, this is very cumbersome for debugging and I haven't found the cause for this yet.

Any ideas what might be causing this? Thanks!

[joelhawksley]

@rockwellll this is the first I've heard of this issue, would you be up for writing a PR with a failing test for this problem?

[rockwellll]

@joelhawksley thanks for the answer! I've did some further investigation, it does not seem to VC's fault per se, rather something from Rails. I'm trying to render a component that has a reference to an undefined variable, and VC correctly reports the error

irb(main):001> WhatsApp::Icon.new.render_in controller.view_context
whatsapp/icon/icon.html.erb:2:in `block in call': undefined local variable or method `hello' for an instance of WhatsApp::Icon (NameError)

When combine this with the render from Rails view helpers, i.e

<%= render WhatsApp::Icon.new %>

The raised error seems to become empty, I'm not sure where things are being rescued from. I checked the render_template_for located in

view_component/lib/view_component/compiler.rb

Lines 62 to 68 in 4ae1dfc

	component_class.silence_redefinition_of_method("render_template_for")
	component_class.class_eval <<-RUBY, __FILE__, __LINE__ + 1
	def render_template_for(variant = nil)
	_call_#{safe_class_name}
	end
	RUBY
	end

But logs seem to become slience at this stage. I'll investigate this further and keep you updated. Is this still related to VC in your thoughts?

[joelhawksley]

Yes, I think this is still a VC problem. For the failing test, please do it in the integration suite ❤️

[rockwellll]

@joelhawksley hi. I just had time to setup what you suggested, and it seems that the issue is on my end. I've added a component that throws an error, VC correctly catches the problem.

I know this is a hard question to guess. But you've any idea what we might be doing to cause this?

[joelhawksley]

I do not. I'd encourage you to continue to try to make a failing test case in your app that isolates the issue. I'd be curious to see what you find!

[rockwellll]

@joelhawksley found the error. It was NewRelic catching the errors reported by VC and swallowed it, I followed this StackOverflow answer which led me to #1981.

Thanks!