This change removes support for the old syntax used for event handlers
and two-way binding.
See the relevant issues for details on the new features and
improvements:
bind https://github.com/aspnet/Blazor/issues/409
event handlers https://github.com/aspnet/Blazor/issues/503
Along with this change we've removed a few additional things Blazor
could do that aren't part of Razor's usual syntax.
----
The features that was used to make something like:
```
<button @onclick(...) />
```
is an expression that's embedded in a an element's attribute. This
feature might be useful in the future if we want to support 'splatting'
arbitrary attributes into a tag, but the runtime support for this isn't
accessible outside the Blazor core.
----
The features that implement:
```
<button onclick=@{ } />
```
have been removed in favor of a better design for lambdas, method group
conversions and other things for event handler attributes.
use `<button onclick=@(x => ...} />` instead.
We think is a better approach in general, because we want the app
developer to write and see the parameter list.
----
Both syntactic features that have been removed have dedicated error
messages in the compiler. If you're porting old code it should help you
figure out what to do.
This isn't needed anymore to support the Blazor design-time experience.
Now that it's gone, it will no longer cause conflicts with MVC's types
so we can remove the other workaround (privateassets).
* Updates the router component to scan for components within assemblies.
* Parses the templates on `[Route]` in component instances and builds a
route table that maps paths to components.
* Uses the route table to map paths to components.
* Switch the tasks used to generate the blazor output to be MSBuild based.
* Package the optimized mono runtime and the BCL inside a nuget package.
* Add opt-in support for linking the application on build.
* Make the whole build process incremental.
This is a working (all tests passing) implementation of the two-phase
compilation system we will need for component discovery.
This builds on top of work we've doing in Razor, including the Razor
SDK, MSBuild tasks, and CLI/server.
This currently *does* discovery components during the build process, but
it doesn't use that data for anything yet.
It works like this:
1. Generate class declarations (structure only, no method bodies)
2. Compile a 'temp' assembly using the .cs files and output of 1.
3. Do component discovery using the 'temp' assembly
4. Generate class definitions (including method bodies)
5. Compile the 'real' assembly using the .cs files and output of 4.
Ryan Nowak
Steve Sanderson
Javier Calvarro Nelson
Ryan Nowak
Eugene Bekker
Dmitry Dizhevskiy