This is a major change on how we represent nested name qualifications in
the AST.
* The nested name specifier itself and how it's stored is changed. The
prefixes for types are handled within the type hierarchy, which makes
canonicalization for them super cheap, no memory allocation required.
Also translating a type into nested name specifier form becomes a no-op.
An identifier is stored as a DependentNameType. The nested name
specifier gains a lightweight handle class, to be used instead of
passing around pointers, which is similar to what is implemented for
TemplateName. There is still one free bit available, and this handle can
be used within a PointerUnion and PointerIntPair, which should keep
bit-packing aficionados happy.
* The ElaboratedType node is removed, all type nodes in which it could
previously apply to can now store the elaborated keyword and name
qualifier, tail allocating when present.
* TagTypes can now point to the exact declaration found when producing
these, as opposed to the previous situation of there only existing one
TagType per entity. This increases the amount of type sugar retained,
and can have several applications, for example in tracking module
ownership, and other tools which care about source file origins, such as
IWYU. These TagTypes are lazily allocated, in order to limit the
increase in AST size.
This patch offers a great performance benefit.
It greatly improves compilation time for
[stdexec](https://github.com/NVIDIA/stdexec). For one datapoint, for
`test_on2.cpp` in that project, which is the slowest compiling test,
this patch improves `-c` compilation time by about 7.2%, with the
`-fsyntax-only` improvement being at ~12%.
This has great results on compile-time-tracker as well:
This patch also further enables other optimziations in the future, and
will reduce the performance impact of template specialization resugaring
when that lands.
It has some other miscelaneous drive-by fixes.
About the review: Yes the patch is huge, sorry about that. Part of the
reason is that I started by the nested name specifier part, before the
ElaboratedType part, but that had a huge performance downside, as
ElaboratedType is a big performance hog. I didn't have the steam to go
back and change the patch after the fact.
There is also a lot of internal API changes, and it made sense to remove
ElaboratedType in one go, versus removing it from one type at a time, as
that would present much more churn to the users. Also, the nested name
specifier having a different API avoids missing changes related to how
prefixes work now, which could make existing code compile but not work.
How to review: The important changes are all in
`clang/include/clang/AST` and `clang/lib/AST`, with also important
changes in `clang/lib/Sema/TreeTransform.h`.
The rest and bulk of the changes are mostly consequences of the changes
in API.
PS: TagType::getDecl is renamed to `getOriginalDecl` in this patch, just
for easier to rebasing. I plan to rename it back after this lands.
Fixes#136624
Fixes https://github.com/llvm/llvm-project/issues/43179
Fixes https://github.com/llvm/llvm-project/issues/68670
Fixes https://github.com/llvm/llvm-project/issues/92757
This patch contains changes for the JSON generator that will enable compatibility with Mustache templates, like booleans to check for the existence and bounds of arrays to avoid duplication.
This patch changes JSON file serialization. Now, files are serialized
to a single directory instead of nesting them based on namespaces. The
global namespace retains the "index.json" name.
This solves the problem of class template specializations being serialized to the
same file as its base template. This is also planned as part of
future integration with the Mustache generator which will consume the JSON files.
Parse friends into a new FriendInfo and serialize them in JSON. We keep track of the friend declaration's template and function information if applicable.
Add support for documenting concepts. This handles concepts and constraints on function and class templates.
Atomic constraints are not considered yet. We don't order constraints based on their conjunctive or disjunctive properties.
LLVM style prefers no default label on fully covered switches to warn if
new enums are added. This patch removes the default label for that
purpose or uses IT_default instead of default if that was the only enum
not covered.
This patch refactors CommentKind handling in clang-doc by introducing a
strongly typed enum class for better type safety and clarity. It updates
all relevant places, including YAML traits and serialization, to work
with the new enum. Additionally, it enhances the Mustache-based HTML
generation by fully supporting all comment kinds, ensuring accurate
structured rendering of comment blocks. The changes simplify future
maintenance, improve robustness by eliminating unchecked defaults, and
ensure consistency between generators.
Fixes https://github.com/llvm/llvm-project/issues/142083
We can only print and use a default arg, if it is instantiated.
I was unable to reduce the test case for this to something of reasonable
size, but this is easily hit by running clang-doc to generate clang's
documentation. For now, we can fix the assertion quickly to unbreak
users, and add the proper test once it is small enough.
Originally part of #133161. This patch adds preliminary tracking
for of TypeInfo, by tracking if the type is a builtin or template.
The new functionality is not yet exercised.
Co-authored-by: Peter Chou <peter.chou@mail.utoronto.ca>
This patch updates Serialize.cpp to serialize more data about C++
templates, which are supported by the new mustache HTML template.
Split from #133161.
Co-authored-by: Peter Chou <peter.chou@mail.utoronto.ca>
Fixes#59812
The names of conversion functions of template type parameters were being
emitted as "type-parameter-N-M". Now we check if the conversion type is
a TemplateTypeParmType and reconstruct the source name.
This patch adds start and end line numbers to clang-doc. Currently
clang-doc only encodes the start line numbers of records, struct, etc.
This patch adds start and end line number to clang-doc bitcode which is
passed to the generator. This will be used by the mustache backend to
generate line ranges. Based on the orginal patch in #135081.
Co-author: Paul Kirth <paulkirth@google.com>
Co-authored-by: PeterChou1 <peter.chou@mail.utoronto.ca>
In quite a few places we were not following the project naming
conventions. This patch applies clang-tidy fixes, and updates
some additional names to follow more typical project wide patterns.
Additionally moving the declarations from Generator.cpp
prevents misc-use-internal-linkage warnings from clang-tidy.
We fix those here too, since the static functions are part
of the same diagnostic
Note that PointerUnion::{is,get} have been soft deprecated in
PointerUnion.h:
// FIXME: Replace the uses of is(), get() and dyn_cast() with
// isa<T>, cast<T> and the llvm::dyn_cast<T>
The implementation in the clang-doc serializer failed to take in the
LangOpts from the declaration. As a result, we'd do things like print
`_Bool` instead of `bool`, even in C++ code.
Fixes#62970
Part of https://github.com/llvm/llvm-project/issues/101129
This patch adds support for attaching comments to enums for HTML in
clang-doc. It changes the enum generation to table tags where as
perviously we're using lists which is more in line with what other doc
generators are doing. It also gives clang-doc the ability to show user
specified enum values
This patch introduces a new enumerator `Invalid = 0`, shifting other enumerators by +1. Contrary to how it might sound, this actually affirms status quo of how this enum is stored in `clang::Decl`:
```
/// If 0, we have not computed the linkage of this declaration.
/// Otherwise, it is the linkage + 1.
mutable unsigned CacheValidAndLinkage : 3;
```
This patch makes debuggers to not be mistaken about enumerator stored in this bit-field. It also converts `clang::Linkage` to a scoped enum.
Reads template information from the AST and adds template parameters and
specialization information to the corresponding clang-doc structures.
Add a "QualName" to the Reference struct which includes the full
qualified type name. The Reference object represents a link in the
HTML/MD generators so is based on the unqualified name. But this does
not encode C-V qualifiers or template information that decorate the
name. The new QualName member encodes all of this information and also
makes it easier for the generators or downsteam YAML consumers to
generate the full name (before they had to process the "Path").
In test code that was changed, remove made-up paths to built-in types
like "int". In addition to slightnly cleaning up the code, these types
do not have paths in real execution, and generating incorrect references
to nonexistant data may complicate future changes in the generators.
Convert llvm::Optional to std::optional (YAML library requires this for
the new usage, and this makes everything consistent according to the
llvm::Optional -> std::optional transition).
Differential Revision: https://reviews.llvm.org/D139154
Reads template information from the AST and adds template parameters and
specialization information to the corresponding clang-doc structures.
Add a "QualName" to the Reference struct which includes the full
qualified type name. The Reference object represents a link in the
HTML/MD generators so is based on the unqualified name. But this does
not encode C-V qualifiers or template information that decorate the
name. The new QualName member encodes all of this information and also
makes it easier for the generators or downsteam YAML consumers to
generate the full name (before they had to process the "Path").
In test code that was changed, remove made-up paths to built-in types
like "int". In addition to slightnly cleaning up the code, these types
do not have paths in real execution, and generating incorrect references
to nonexistant data may complicate future changes in the generators.
Differential Revision: https://reviews.llvm.org/D139154
Read typedef and "using" type alias declarations and serialize into the
internal structures. Emit this information in the YAML output. The HTML
and MD generators are unchanged.
Separate out the logic to create the parent namespace or record object
and insert the newly created child into it. This logic was previously
duplicated for every "info" type and is now shared.
To help this, a struct containing the child vectors was separated out so
children can be added generically and without having too many templates.
A small change was made to populateParentNamespaces() to allow using
types that aren't themselves DeclContexts (typedefs are the first
example of this).
Differential Revision: https://reviews.llvm.org/D134371
Read typedef and "using" type alias declarations and serialize into the internal structures. Emit this information in the YAML output. The HTML and MD generators are unchanged.
Separate out the logic to create the parent namespace or record object and insert the newly created child into it. This logic was previously duplicated for every "info" type and is now shared.
To help this, a struct containing the child vectors was separated out so children can be added generically and without having too many templates.
A small change was made to populateParentNamespaces() to allow using types that aren't themselves DeclContexts (typedefs are the first example of this).
Reviewed By: paulkirth, haowei
Differential Revision: https://reviews.llvm.org/D134371
The *Info object (for the copy of the AST") constructors had many duplicated variants. Many of the variants seemed to be in an attempt to avoid default arguments. But default arguments are not prohibited and using them allows most of the variants to be removed which improves readability.
Remove the IsInGlobalNamespace flag on a Reference. This is set when the path is empty, and only read once in the HTML generator with the identical condition. The constructor cleanup exposed a problem where this was set to false when the constructor with no path was used, but true when the path was set to empty.
There should be no observable change with the exception that IsInGlobalNamespace is no longer emitted in YAML.
Reviewed By: paulkirth, haowei
Differential Revision: https://reviews.llvm.org/D134235
Several different places in the code had similar computations for the parameters that were eventually passed to the TypeInfo constructor.
This centralizes that code in one function, and allows passing TypeInfo to the various other *Info structures that need it.
Remove some "auto" types and replace with the real type for getting declarations. This was making some duplicate checking difficult to see.
Reviewed By: paulkirth
Differential Revision: https://reviews.llvm.org/D134225
Add support for explicitly typed enums:
enum Foo : unsigned { ... };
to the internal representation and to the YAML output.
Add support for getting the value of an enum constant, as well as accessing the original expression that produced it. This changes the YAML output of enums from an array of strings for the enum members to an array of dictionaries. These dictionaries now report the name, value, and original expression.
The markdown and HTML outputs are unchanged, they still output the name from the new enhanced internal schema.
Reviewed By: paulkirth
Differential Revision: https://reviews.llvm.org/D134055
Adds support for default arguments in the internal representation and reads these values from the source. Implements writing these values to YAML but does not implement this for the HTML or markdown outputs.
Reviewed By: paulkirth
Differential Revision: https://reviews.llvm.org/D133732
Struct/class data members did not have the comments associated with
them. This adds that information to the MemberTypeInfo class and emits
it in the YAML. This does not update the frontends yet.
Reviewed By: paulkirth
Differential Revision: https://reviews.llvm.org/D131298
The pointer is referenced immediately, so assert the cast is correct instead of returning nullptr
It's only later iterations of the loop where the getParent() call might return nullptr
When a Record is declared in the global namespace, clang-doc serializes
it as a child of the global namespace, so the global namespace is now
one if its parent namespaces. This namespace was not being included in
the list of namespaces of the Info causing paths to be incorrect and the
index rendered incorrectly.
Affected tests have been fixed.
Differential revision: https://reviews.llvm.org/D66298
llvm-svn: 369123
clang-doc now serializes the inherited attributes and methods, not only the name of the base class.
All inherited are tracked, if B:A and C:B, info of A is included in C.
This data is stored in attribute Bases in a RecordInfo.
Previously tracked inheritance data, stored in Parents and VParents, hasn't been removed to reduce review load.
Differential revision: https://reviews.llvm.org/D66238
llvm-svn: 369075
Bitcode writer was not emitting the corresponding record for the Access attribute of a FunctionInfo. This has been added.
AS_none was being used as the default value for any AcesssSpecifier attribute
(in FunctionInfo and MemberTypeInfo), this has been changed to AS_public
because this is the enum value that evaluates to 0.
The bitcode writer doesn't write values that are 0 so if an attribute
was set to AS_public, this value is not written and after reading the
bitcode it would have the default value which is AS_none. This is why
the default value is now AS_public.
Differential Revision: https://reviews.llvm.org/D66151
llvm-svn: 369063
Now that we've moved to C++14, we no longer need the llvm::make_unique
implementation from STLExtras.h. This patch is a mechanical replacement
of (hopefully) all the llvm::make_unique instances across the monorepo.
Differential revision: https://reviews.llvm.org/D66259
llvm-svn: 368944
Path is now stored in the references to the child while serializing,
then this path is used to generate the relative path in the HTML
generator.
Now some references have paths and some don't so in the reducing phase,
references are now properly merged checking for empty attributes.
Tests added for HTML and YAML generators, merging and serializing.
computeRelativePath function had a bug when the filepath is part of the
given directory; it returned a path that starts with a separator. This
has been fixed.
Differential Revision: https://reviews.llvm.org/D65987
llvm-svn: 368602
Two command line options have been added to clang-doc.
--repository=<string> - URL of repository that hosts code; used for links to definition locations.
--source-root=<string> - Directory where processed files are stored. Links to definition locations will only be generated if the file is in this dir.
If the file is in the source-root and a repository options is passed;
a link to the source code will be rendered by the HTML generator.
Differential Revision: https://reviews.llvm.org/D65483
llvm-svn: 368460
<a> tags are added for the parents and members of records and return type and
params of functions. The link redirects to the reference's info file.
The directory path where each info file will be saved is now generated in the
serialization phase and stored as an attribute in each Info.
Bitcode writer and reader were modified to handle the new attributes.
Committed on behalf of Diego Astiazarán (diegoaat97@gmail.com).
Differential Revision: https://reviews.llvm.org/D63663
llvm-svn: 365937
Serialization of child namespaces and records is now handled.
Namespaces can have child records and child namespaces.
Records can only have child records.
Committed on behalf of Diego Astiazarán (diegoaat97@gmail.com).
Differential Revision: https://reviews.llvm.org/D63911
llvm-svn: 364963
