This change splits up the conventional routing path from the attribute
routing path *inside* routing, instead of inside `MvcRouteHandler`. Each
attribute route group now gets its own instance of
`MvcAttributeRouteHandler` which just knows about the actions it can
reach.
This removes the concept of a route-group-token and removes the lookup
table entirely for attribute routing. This also means that the
`DefaultHandler` on `IRouteBuilder` will not be used for attribute routes,
which we are OK with for 1.0.0.
The action selector's functionality is now split into two methods. We
think this is OK for 1.0.0 because any customization of `IActionSelector`
up to now had to implement virtually the same policy as ours in order to
work with attribute routing. It should now be possible to customize the
selector in a meaningful way without interfering with attribute routing.
Believe it or not, this one in ActionResult is showing in TechEmpower. I
did a pass on all of the product code since the cache task actually
improves the readability of the code 😎
Just a safety measure in case anything evades our 'flattening'. I don't
have any evidence that we're failing to clear something, but we want this
here as a safeguard.
- Simplify things that used to rely on HtmlTextWriter. This behavior is
now the default.
- Simplify a whole mess of Razor TextWriter code.
- Integration test for merging of TagHelper buffers back into the 'main'
buffer.
This change removes a layer of abstraction. These validators now just do
what they do now without creating a bunch of intermediate objects.
ModelClientValidationRule has been removed, and client validations
manipulate the attributes collection of a tag directly.
These changes are aimed at significantly improving the performance of
MVC/Razor when a large amount of content is in play or a large number of
TagHelpers are used.
A few issues addressed:
- Buffer sync writes after a flush has occurred so that we can write them
asyncronously. The issue is that an IHtmlContent can only do sync
writes. This is very bad for Kestrel in general. Doing these writes
async is better for our overall perf, and the buffer that we use for it
is from the pool.
- 'Flatten' ViewBuffers when possible. A page with lots of TagHelpers can
end up renting a ViewBuffer and only write 2-3 things into it. When a
ViewBuffer sees another ViewBuffer we can either steal its pages, or
copy data out and 'return' its pages. This lets us use 3-4 buffers for a
large Razor page instead of hundreds.
Ryan Nowak
jacalvar
Doug Bunting
BrennanConroy
ryanbrandenburg
mnltejaswini
Pranav K
Ajay Bhargav Baaskaran
N. Taylor Mullen