-
Notifications
You must be signed in to change notification settings - Fork 10.5k
Front-end: add a new module loader that loads explicitly built Swift modules #32170
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
e663888
to
be687b1
Compare
@swift-ci please smoke test |
…modules To support -disable-implicit-swift-modules, the explicitly built modules are passed down as compiler arguments. We need this new module loader to handle these modules. This patch also stops ModuleInterfaceLoader from building module from interface when -disable-implicit-swift-modules is set.
be687b1
to
bd782be
Compare
@swift-ci please smoke test |
1 similar comment
@swift-ci please smoke test |
@@ -2190,6 +2190,30 @@ TypeDecl *ModuleFile::lookupLocalType(StringRef MangledName) { | |||
return cast<TypeDecl>(getDecl(*iter)); | |||
} | |||
|
|||
std::unique_ptr<llvm::MemoryBuffer> | |||
ModuleFile::getModuleName(ASTContext &Ctx, StringRef modulePath, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why do we need to load the module here? Can we instead provide more information on the command line, e.g., a specific mapping from module name to the module file the contains it?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah, we could do that to avoid loading modules eagerly. How about we accept a JSON file from the command line to include module name, module path, swiftdoc path and swiftsourceinfo path?
@@ -450,6 +450,16 @@ bool CompilerInstance::setUpModuleLoaders() { | |||
return true; | |||
} | |||
|
|||
// If implicit modules are disabled, we need to install an explicit module | |||
// loader. | |||
if (Invocation.getFrontendOptions().DisableImplicitModules) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
How about modeling DisableImplicitModules
as a ModuleLoadingMode
? It fits in with the scheme established there.
} | ||
// We found an explicit module matches the given name, give the buffer | ||
// back to the caller side. | ||
*ModuleBuffer = std::move(it->getValue().moduleBuffer); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We also need to deal with the .swiftdoc
and .swiftsourceinfo
as well, right?
Invocation.getSearchPathOptions().ExplicitSwiftModules, | ||
IgnoreSourceInfoFile); | ||
Context->addModuleLoader(std::move(ESML)); | ||
} | ||
if (MLM != ModuleLoadingMode::OnlySerialized) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
When ExplicitSwiftModuleLoader
is added, it seems like we should not add ModuleInterfaceLoader
or SerializedModuleLoader
at all. Then we don't need checks for DisableImplicitModules
anywhere.
To support -disable-implicit-swift-modules, the explicitly built modules
are passed down as compiler arguments. We need this new module loader to
handle these modules.
This patch also stops ModuleInterfaceLoader from building module from interface
when -disable-implicit-swift-modules is set.