This patch adds frames emitting for
`bugprone-exception-escape.TreatFunctionsWithoutSpecificationAsThrowing`.
As of AI Usage: Gemini 3 is used for pre-commit reviewing.
Closes https://github.com/llvm/llvm-project/issues/184781
`BindingDecl` nodes, i.e. the individual names in a structured binding,
were not handled in `IdentifierNamingCheck::findStyleKind()`, causing
them to fall through to the Default style or be silently ignored.
This led to incorrect renames, e.g. applying member variable conventions
to local bindings.
---------
Signed-off-by: Björn Svensson <bjorn.a.svensson@est.tech>
When filtering is enabled, its possible an Info doesn't have a
Parent USR. Use `find()` to safely handle that case.
Additionally, I noticed the comparison code for the index
poorly reimplemented the existing comparison from StringRef.
We can just use the one from ADT.
The `run-clang-tidy.py` script now uses `shlex.join()` to construct the
command string for printing.
This ensures that arguments containing shell metacharacters, such as the
asterisk in `--warnings-as-errors=*`, are correctly quoted. This allows
the command to be safely copied and pasted into any shell for manual
execution, fixing errors previously seen with shells like `fish` that
are strict about wildcard expansion.
Before:
```
[ 1/15][0.2s] /usr/bin/clang-tidy -p=/home/user/work/project/build --warnings-as-errors=* /home/user/work/project/src/main.cpp
```
Note: When running this command in fish shell you get some error like
`fish: No matches for wildcard '--warnings-as-errors=*'. See `help
wildcards-globbing``
After:
```
[ 1/15][0.2s] /usr/bin/clang-tidy -p=/home/user/work/project/build '--warnings-as-errors=*' /home/user/work/project/src/main.cpp
```
Combine the two separate test files and have them feed from a common
source. This will be the way that tests are handled to prevent testing
divergence in the future.
Since stddef.h in the mock headers of clang-tidy tests now provides
`#define NULL 0L`, we can migrate manual `#define NULL` in these tests
to standard `#include`s.
Added missing ``#include`` insertion when the format function call
appears as an argument to a macro.
Part of #175183
---------
Co-authored-by: Victor Chernyakin <chernyakin.victor.j@outlook.com>
Do not provide diagnostics for cert-dcl58-cpp for compiler generated
intrinsic as it will be a false positive.
In provided tests compiler generates align_val_t which ends up inside
std namespace, resulting in std::align_val_t symbol. This symbol is
compiler generated, having no location, causing compiler crash. Also
there is no point to notify user about violations which user has no
control of.
Resolution: Diagnostics suppressed.
Co-authored-by: Vladislav Aranov <vladislav.aranov@ericsson.com>
Introduce a standalone tool to remap paths inside clangd background
index files. This allows users to pay the cost of generating a full
background index once, then reuse the background index across multiple
clients. Each client rewrites the background index in place after
copying the original, specifying the path(s) to be remapped.
This is an alternative approach to the dynamic path remapping proposed
in #180285.
Fixesclangd/clangd#847
Assisted-by: claude
To address clangd maintance problem, I'm nominating myself as a clangd
maintainer.
I'm currently actively reviewing PRs. I also participated in
clangd-related discourse discussions, subscribed to clangd issues and
PRs. For a while I plan to spend some time for PRs review in clangd.
## Summary
This change adds `unless(isInstantiationDependent())` to the
`ArraySubscript` matcher.
## Motivation
In template code, some array subscript expressions are not fully
resolved yet.
Because of this, the matcher may treat them incorrectly and produce
false positives.
## Change
This patch skips instantiation-dependent expressions in templates.
It helps the matcher work only on resolved expressions.
Closes https://github.com/llvm/llvm-project/issues/187009.
This commit fixes a false positive in the
derived-method-shadowin-base-method clang-tidy check, as described in
[ticket 185741](https://github.com/llvm/llvm-project/issues/185741)
Fixes#185741
---------
Co-authored-by: Tom James <tom.james@siemens.com>
Co-authored-by: Zeyi Xu <mitchell.xu2@gmail.com>
…ule files
When compile_commands.json contains relative paths in -fmodule-file=
arguments (as generated by CMake), clangd failed to find the BMI files
because it was looking for them relative to the wrong working directory.
This patch fixes the issue by converting relative paths to absolute
paths based on the compilation directory (CompileCommand.Directory)
before checking if the module file exists and is up to date.
Added a unit test that verifies the fix works correctly.
AI Assisted
The IsModuleFileUpToDate function was not properly validating input
files for C++20 modules. By default, ASTReader skips input file
validation for StandardCXXModule files unless
ForceCheckCXX20ModulesInputFiles and ValidateASTInputFilesContent are
both set.
This change:
- Passes ValidateASTInputFilesContent=true to ASTReader constructor
- Uses ARR_OutOfDate flag for cleaner error handling
- Simplifies the validation logic (ReadAST already validates internally)
- Adds a test to verify header changes in module units are detected
Assised with AI.
About half of this check's code is dedicated to implementing a pair of
set containers with optimizations for when the element count is small.
But the check only uses these containers while constructing the warning
message. That's not generally a hot path in any check. Just to confirm,
I ran the check over `clang/lib/Sema/Sema.cpp` and all its included
headers before and after, and saw no performance difference. So, these
containers seem like a false optimization.
Code completion previously could not tell apart the declaration of
a method from a call to it, and provided call-like behaviour even
in declaration contexts. This included things like not offering
completion for private methods, and inserting placeholders for
the parameters as though the user was going to fill in arguments.
This patch adds support to Parser and SemaCodeComplete to
detect and provide dedicated behaviour for declaration contexts.
In these contexts, the flag CodeCompletionResult::DeclaringEntity
is set, and createCodeCompletionString() is adjusted to handle this
flag, e.g. by inserting parameter declarations as text chunks rather
than placeholder chunks.
The DeclaringEntity flag is also available for consumers of
SemaCodeComplete, such as clangd, to customize their behaviour.
In addition, the patch tweaks the conditions under which the
existing CodeCompletionResult::FunctionCanBeCall flag is set to
be more accurate, excluding declaration contexts and cases where
the address of the function is likely being taken.
Fixesclangd/clangd#753Fixesclangd/clangd#880Fixesclangd/clangd#1752
Extend the bugprone-use-after-move check to recognize user defined
smart-pointer-like types that make guarantees on the state of a
moved-from object, leaving it in a valid and specified state that
matches the standard smart pointer's moved-from state (nullptr), where
it is safe to use but not dereference.
Following the RFC discussion:
* Use `[[clang::annotate]]` to mark the types.
* Use an schema for the `[[clang::annotate]]` annotation and arguments
to help avoid conflicts with other users of the attribute.
* The annotation will identify the tool ("clang-tidy") and the arguments
the plugin ("bugprone-use-after-move"), and the behavior of the type
("null_after_move"). E.g.:
`[[clang::annotate("clang-tidy", "bugprone-use-after-move",
"null_after_move")]]`
RFC:
https://discourse.llvm.org/t/rfc-add-a-class-attribute-clang-null-after-move-for-use-after-move-analysis/89760
```txt
---User Time--- --System Time-- --User+System-- ---Wall Time--- --- Name ---
Status quo: 0.4743 (100.0%) 0.3802 (100.0%) 0.8546 (100.0%) 0.8567 (100.0%) bugprone-suspicious-semicolon
With this change: 0.0103 (100.0%) 0.0027 (100.0%) 0.0130 (100.0%) 0.0133 (100.0%) bugprone-suspicious-semicolon
```
Continuing the trend of registering one `anyOf` matcher being slower
than registering each of its matchers separately (see #178829 for a
previous example).
(This PR also changes the traversal mode, but I only saw a small speedup
from that. Most of it came from registering the matchers separately.)
This check wasn't super expensive to begin with, but the speedup is
still pretty nice.
Function template specialization arguments were incorrectly rendered
without a comma. This was due to the "End" JSON property also being
used in the levels above. Mustache looks for missing properties in
parent contexts, see #174359.
The `check_alphabetical_order.py` script previously only scanned the
first line of each bullet point in `ReleaseNotes.rst`, causing sorting
failures when a `:doc:` tag was split across multiple lines.
Also, when it is sorting the last entry of a section, the script will
insert an unnecessary whitespace.
This PR fixes these two problems.
This removes the assumption that a deserialized module is backed by a
`FileEntry`. The `OptionalFileEntryRef` member is replaced with
`ModuleFile{Name,Key}`.
