Deduplicating CanReadRequestBody code.

content/browser/loader/DEPS

 # The core loading code in this directory is being transitioned into a separate
 # Mojo networking service. See http://crbug.com/598073. The first stage is to
 # decouple the code that will be moved from the rest of content. These rules
 # help ensure that we don't add more includes, and can progressively reduce the
 # code's dependencies.
 #
 # For each file, the first section shows includes that are allowed (i.e. because
 # they are for files that will be moving as well). The second section, separated
 # by a comment, lists files that we will have to decouple.
 specific_include_rules = {
@@ skipping 136 matching lines @@
     "+content/browser/loader/cross_site_resource_handler.h",
     "+content/browser/loader/detachable_resource_handler.h",
     "+content/browser/loader/intercepting_resource_handler.h",
     "+content/browser/loader/mime_sniffing_resource_handler.h",
     "+content/browser/loader/navigation_resource_handler.h",
     "+content/browser/loader/navigation_resource_throttle.h",
     "+content/browser/loader/navigation_url_loader_impl_core.h",
     "+content/browser/loader/redirect_to_file_resource_handler.h",
     "+content/browser/loader/upload_data_stream_builder.h",
     "+content/browser/resource_context_impl.h",
+    "+content/browser/resource_request_body_browser_utils.h",

[Łukasz Anforowicz, 2017/05/25 16:11:37]

I assume that after decoupling from the rest of //content,
content/browser/loader will define an abstract LoaderDelegate interface that
will be used to interact with things that should stay in the //content.  I
assume that ResourceDispatcherHostImpl::ShouldServiceRequest(...) will move into
LoaderDelegate::ShouldServiceRequest(...) that will overriden/implemented in the
content layer.  Therefore I think it is okay to add the dependency above - this
dependency will stay in the //content layer in the end (i.e. the dependency
would be used from ContentLoaderDelegate::ShouldServiceRequest).

Does what I said above make sense?

How can we deduplicate the code while still supporting the efforts for
decoupling the loader from the rest of //content?

[mmenke, 2017/05/25 16:31:08]

For better or for worse, the plan is now to make a service from scratch, so it
won't be using this code at all.

The RDH-equivalent will be in a separate process, unclear how much, if any,
Chrome code will be in that process.

[Łukasz Anforowicz, 2017/05/25 16:50:26]

Does that mean that the change to the DEPS file looks okay?

     "+content/browser/service_worker/foreign_fetch_request_handler.h",
     "+content/browser/service_worker/link_header_support.h",
     "+content/browser/service_worker/service_worker_context_wrapper.h",
     "+content/browser/service_worker/service_worker_navigation_handle_core.h",
     "+content/browser/service_worker/service_worker_request_handler.h",
     "+content/browser/streams/stream.h",
     "+content/browser/streams/stream_context.h",
     "+content/browser/streams/stream_registry.h",
     "+content/common/content_export.h",
     "+content/common/net/url_request_service_worker_data.h",
@@ skipping 178 matching lines @@
     "+content/browser/loader/navigation_resource_throttle.h"
   ],
   "wake_lock_resource_throttle\.(cc|h)": [
     "-content",
     "+content/browser/loader/wake_lock_resource_throttle.h",
     "+content/browser/service_manager/service_manager_context.h",
     "+content/public/browser/browser_thread.h",
     "+content/public/browser/resource_throttle.h",
   ],
 }