Use page refreshes

[jorgemanrubia]

No description provided.

[northeastprince]

@jorgemanrubia is the refresh broadcast working for you? It wasn't for me, which is why I opened hotwired/turbo-rails#522.

[yshmarov]

does Hey really use callbacks/touch like this in real apps, or is it only for the sake of example?

[northeastprince]

Yeah, they most likely do, since it's DRY and allows you to quickly invalidate caches as well, which they use heavily.

[jorgemanrubia]

Yes, I can confirm we do!

[jorgemanrubia]

@northeastprince it's working indeed. Do you mean that it's not working even if you clone this repo? Or in your app? If the later, which Rails version are you using?

[northeastprince]

@jorgemanrubia I shared the issue here, my app is running on Rails edge. Let me know if you need more info!

[bradgessler]

I'm experimenting with Turbo 8 via Phlex components and was surprised to see that the meta tags all use the provide :head construct instead of just emitting a tag.

If you want to make the tags accessible from outside of the content view, you might either eliminate the provide :head construct or keep that and expand out the methods like this:

  # Pages that are more likely than not to be a cache miss can skip turbo cache to avoid visual jitter.
  # Cannot be used along with +turbo_exempts_page_from_preview+.
  def turbo_exempts_page_from_cache
    provide :head, turbo_exempts_page_from_cache_tag
  end

  def turbo_exempts_page_from_cache_tag
    tag.meta(name: "turbo-cache-control", content: "no-cache")
  end

I think the reason you want to keep the provide :head tag is because the meta tags within each view file needs to put the meta tag in the head? The approach I provided above should serve both needs.

[bradgessler]

Here's what I ended up doing in my Phlex layout:

        # Turbo 8 has the helper method `turbo_refreshes_with method: :morph, scroll: :preserve` that
        # inserts the content in the `head` tag. Since Phlex doesn't work that way, we'll just put the tags here.
        meta(name: "turbo-refresh-method", content: "morph")
        meta(name: "turbo-refresh-scroll", content: "preserve")

Not a huge deal, but it would be nice to go through the logic that checks for argument validity if the tags are broken apart.

[krschacht]

@bradgessler from Chicago / Poll Everywhere. Funny running into you on a random PR comment thread. :)

[bradgessler]

@krschacht speaking to you live from CA, and running into similar ergonomic issues that you're running into.

[bradgessler]

I opened a PR at hotwired/turbo-rails#542 that decouples tag generation from the way it's served. We can continue this discussion over there.

[krschacht]

@jorgemanrubia When I open the projects index (http://localhost:3000/projects) in two separate browser windows, and in one of them I create a new project, the other one does not update with the new project. I have to manually refresh the page to see it.

I assume it's because of this block here ^, it only subscribes to the specific projects which exist at the time the page loads so it's not subscribed to a newly created project. One solution would be to introduce a new higher-level abstraction called Portfolio (a collection of projects) and broadcast refreshes from that. But that seems heavy-handed if we don't otherwise need a new model.

Is there a simpler solution to this within the Turbo 8 Morphing world?

[jorgemanrubia]

Yes. Check this PR hotwired/turbo-rails#521. You can use the pluralized model name as the stream name for listening to additions.

[martijn10kb]

But would this also work when we have scopes in place? For example like:

  <%= turbo_stream_from @projects.not_deleted.is_active.where(client_id: 1) %>

Or would it just subscribe to all the projects?