Supply "IsConnected" state info to components #8888
Conversation
| // Copyright (c) .NET Foundation. All rights reserved. | ||
| // Licensed under the Apache License, Version 2.0. See License.txt in the project root for license information. | ||
|
|
||
| namespace Microsoft.AspNetCore.Components.Services |
There was a problem hiding this comment.
What would you think about putting this in Microsoft.AspNetCore.Components. I really think we'll end up removing .Services as a namespace when this gets reviewed in favor of some mix of:
- moving these types into
.Components - moving these types into more specific namespaces like routing/layouts
- moving them into a more generic namespace like
.Infrastructurefor things that most people won't use
There was a problem hiding this comment.
Totally happy to merge the Components.Services namespace into .Components. That would also fix one of the other things that's been bothering me lately - that the templates imports don't include .Services.
Filed #8892
| /// <example>During prerendering, the value will always be false.</example> | ||
| /// <example>During server-side execution, the value can be true or false depending on whether there is an active SignalR connection.</example> | ||
| /// <example>During client-side execution, the value will always be true.</example> | ||
| bool IsConnected { get; } |
There was a problem hiding this comment.
Should we document some guarantees about what happens when transition between states? Or is this not the right place for that.
There was a problem hiding this comment.
What kind of guarantees do you have in mind?
There was a problem hiding this comment.
Like, if you weren't connected when you last rendered - does that mean that you get called again when you are connected?
There was a problem hiding this comment.
Ah, I see. Since we're not proposing to do that, I'm not sure what there would be to say about it here.
I expect you know already, but the intent to deal with the mainstream "transition from prerendered to interactive" case is via lifecycle methods. For a more general "notify me when connection state changes", I'd expect that to be a separate API, which someone could wrap into a cascading value if they wanted.
There was a problem hiding this comment.
Ok cool. I guess what I was thinking about is where are we going to document to the pattern for using this property effectively. That's probably not a good fit for the in-code docs, so feel free to ignore.
Eilon
added
the
area-mvc
SteveSandersonMS
force-pushed
the
stevesa/icomponentcontext
branch
from
69dbebe to
6df3556
Compare
|
I already included something similar in my draft PR :( |
| endpoints.MapFallbackToPage("/PrerenderedHost"); | ||
| endpoints.MapComponentHub<TestRouter>("app"); | ||
| }); | ||
| }); |
There was a problem hiding this comment.
This should not be our experience for having multiple entry points as part of an app. But that requires a separate discussion. See #8913
There was a problem hiding this comment.
Sure, this is just what you have to do today, not a claim that it's the ideal dev experience or that we shouldn't change it.
| @@ -836,6 +836,10 @@ public enum NavLinkMatch | |||
| } | |||
| namespace Microsoft.AspNetCore.Components.Services | |||
| { | |||
| public partial interface IComponentContext | |||
There was a problem hiding this comment.
How did you update this? Can you try and get the private members of the components that we ship to show up? That breaks the current E2E test in ProjectTemplates. It would be incredibly dope if we updated and enable that.
There was a problem hiding this comment.
I ran GenerateReferenceSource as per https://github.com/aspnet/AspNetCore/blob/master/docs/ReferenceAssemblies.md
What you see here is what it outputs. You can't make it output private members as they are not an aspect of reference assemblies.
There was a problem hiding this comment.
I do believe we did this for some other APIs in the assembly. This is an ongoing problem on the templates. Things like Assembly="typeof(Blah).Assembly" won't work because it won't be able to find the Assembly property on the component and it will treat it as a string instead when compiling.
There was a problem hiding this comment.
Yes, @rynowak fixed that by telling the tool not to generate anything for those types, and he wrote those ref types manually. I don't think we need to do IComponentContext manually though.
There was a problem hiding this comment.
For now we need to manually generate the reference source of components because non-public methods do not show up in reference assemblies. This should not apply to anything else we need to do. If the contagion starts spreading I would love to learn more about that 😆
SteveSandersonMS
force-pushed
the
stevesa/icomponentcontext
branch
from
6df3556 to
edf0347
Compare
This is a partial implementation of what we need for #8786. It's pretty basic and just follows from existing patterns.
Still to do: E2E tests showing the swap from "prerendering/disconnected" into "connected", with a component that renders different output in the two cases.
Question: Do we need some other flag to differentiate the "prerendering" case from the "was interactive but then disconnected" case? I'm not sure we have a use case for that.