LibRequests+LibWeb+RequestServer: Propagate HTTP reason phrase

[rmg-x]

Fixes 54 test cases across the following WPT tests :^)

• http://wpt.live/xhr/status-basic.htm
• http://wpt.live/xhr/status-async.htm

[rmg-x]

Marking this ready for review -- I don't see a good way of adding in-tree tests for this without something like an echo server proposed in #1801

[ladybird-bot]

Hello!

One or more of the commit messages in this PR do not match the Ladybird code submission policy, please check the lint_commits CI job for more details on which commits were flagged and why.
Please do not close this PR and open another, instead modify your commit message(s) with git commit --amend and force push those changes to update this PR.

[alimpfard]

LibWeb LibWebView

Not exactly your fault but...why does RS need to link against anything libweb?

[alimpfard]

This really goes way back: 58e3b8c (though it makes sense in a weird way)
Also cc @trflynn89...that seems like a hack at best? why not depend directly on generate_RequestClientEndpoint.h and generate_RequestServerEndpoint.h?

[trflynn89]

We used to have a really hacky way to get IPC-generated files into the ladybird build, back when we had to deal with both lagom and serenity sources:

ladybird/Meta/Lagom/CMakeLists.txt

Lines 425 to 441 in 58e3b8c

	if (ENABLE_LAGOM_LIBWEB)
	list(APPEND lagom_standard_libraries Web)
	# WebView
	list(APPEND LIBWEBVIEW_SOURCES "../../Userland/Libraries/LibWebView/AccessibilityTreeModel.cpp")
	list(APPEND LIBWEBVIEW_SOURCES "../../Userland/Libraries/LibWebView/DOMTreeModel.cpp")
	list(APPEND LIBWEBVIEW_SOURCES "../../Userland/Libraries/LibWebView/StylePropertiesModel.cpp")
	list(APPEND LIBWEBVIEW_SOURCES "../../Userland/Libraries/LibWebView/ViewImplementation.cpp")
	list(APPEND LIBWEBVIEW_SOURCES "../../Userland/Libraries/LibWebView/WebContentClient.cpp")
	list(APPEND LIBWEBVIEW_SOURCES "../../Userland/Libraries/LibWebView/RequestServerAdapter.cpp")
	compile_ipc(${SERENITY_PROJECT_ROOT}/Userland/Services/WebContent/WebContentServer.ipc WebContent/WebContentServerEndpoint.h)
	compile_ipc(${SERENITY_PROJECT_ROOT}/Userland/Services/WebContent/WebContentClient.ipc WebContent/WebContentClientEndpoint.h)
	compile_ipc(${SERENITY_PROJECT_ROOT}/Userland/Services/WebContent/WebDriverClient.ipc WebContent/WebDriverClientEndpoint.h)
	compile_ipc(${SERENITY_PROJECT_ROOT}/Userland/Services/WebContent/WebDriverServer.ipc WebContent/WebDriverServerEndpoint.h)
	compile_ipc(${SERENITY_PROJECT_ROOT}/Userland/Services/RequestServer/RequestClient.ipc RequestServer/RequestClientEndpoint.h)
	compile_ipc(${SERENITY_PROJECT_ROOT}/Userland/Services/RequestServer/RequestServer.ipc RequestServer/RequestServerEndpoint.h)

Don't know if we're still doing this, but I can't imagine there's a reason we need to continue that hack

[alimpfard]

We're still doing that (roughly, those are listed in the GENERATED_SOURCES for WebView), but even then there's no real reason to actually link that thing in afaict

[ADKaster]

In the gn build we generate the header in the client library build script. I would prefer that here in CMake as well.

[alimpfard]

Do we really need to link against libweb just to get a stringbuilder + append_code_point in a loop?
Also cc @trflynn89 (iirc you merged that), maybe we could put that function in the header or just inline here?

[trflynn89]

It could also be moved to LibTextCodec

[rmg-x]

This PR was previously linking against LibTextCodec and just using a Latin1 decoder directly. Should I just revert to using that again?

[alimpfard]

Alright, let's get this in the tree, we can fix the dependency problem later.