| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
| |
Wasn't warning for:
QStringList list = { "foo" };
The child CXXConstructExprs appear in the AST but are never visited.
|
| |
|
|
|
|
|
|
|
|
|
| |
fixes --list-checks
I can't reproduce the failure anymore so removing the llvm-9
workarounds
CCMAIL: kfunk@kde.org
|
|
|
|
| |
CCMAIL: kfunk@kde.org
|
|
|
|
|
|
| |
std::isspace is part of <cctype>
REVIEW: D24513
|
|
|
|
| |
MSVC doesn't know about 'uint'
|
|
|
|
| |
Prior to version 9 the last character was missing in the macro
|
|
|
|
| |
Before llvm-9 the last character was missing.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is a temporary fix so it doesn't block distros from using clazy
with llvm-9. And I'm not even sure others can reproduce, as it
might be a local compiler bug here. Debugged it quite a bit and
couldn't reach any conclusion.
Nikolai, it would be great if you could run:
$ cd tests/
$ clazy-standalone auto-unexpected-qstringbuilder/main.cpp
after having uncommented:
// llvm::errs() << optionsParser.getSourcePathList().size() << "\n";
in ClazyStandaloneMain.cpp
should not print 0
CCMAIL: nikolai.kosjar@qt.io
|
| |
|
|
|
|
| |
Typedefs from system headers (such as Qt) were being ignored
|
|
|
|
| |
As some checks need to know about all typedefs
|
| |
|
| |
|
| |
|
|
|
|
|
| |
These don't indicate bugs and are common place.
We can maybe wrap them with an option to enable them.
|
|
|
|
|
|
|
| |
having more than 1 param is commonplace and not a bug.
Although elegant to only have 1, if we warn we'll produce lots of noise
and people will probably disable this check instead, failing to find more
important bugs
|
| |
|
|
|
|
| |
Type comparison is made in several places, so add it to a function
|
|
|
|
| |
When using inner classes as Q_PROPERTY
|
|
|
|
|
| |
It's common place for the NOTIFY signal to not have arguments, and
there's no performance improvement on not adding one
|
|
|
|
|
| |
Namely enums and bool properties. Enums had the enum tag
when printing, and bool was Bool_.
|
|
|
|
|
| |
It's a bit of an hack and very specific to range-loop, not something
to be reused
|
|
|
|
|
|
|
|
| |
This now warns:
QStringList list = getList();
for (const QString &s : list)
CCMAIL: dfaure@kde.org
|
| |
|
|
|
|
|
|
| |
Summary: This is in line with the same option of clang-tidy.
Differential Revision: https://phabricator.kde.org/D23872
|
|
|
|
|
|
| |
Reviewers: smartins
Differential Revision: https://phabricator.kde.org/D23871
|
|
|
|
|
|
|
|
|
|
|
|
| |
Summary:
...to dump the contents of checks.json.
This is useful for IDEs as they can find out now whether a particular
check is supported.
BUG: 409986
Differential Revision: https://phabricator.kde.org/D23680
|
|
|
|
| |
has many false positives for now
|
| |
|
|
|
|
|
|
|
| |
Warns when any of the functions or variables declared in a Q_PROPERTY
have types, arguments or return types differing with the Q_PROPERTY.
REVIEW: D22962
|
|
|
|
|
|
| |
The message didn't have the [checkName] tags
BUG: 411440
|
|
|
|
|
|
|
| |
TUDiag was being exported multiple times rewriting the same file.
Export at the end only.
BUG: 409983
|
| |
|
|
|
|
| |
Just to dump records and translation units-for now.
|
| |
|
|
|
|
| |
For the same reason as return values
|
|
|
|
| |
It's a design smell to have signals returning values
|
|
|
|
|
|
|
|
|
| |
Instead of saying that no cast to base is needed, simply say that
no qobject_cast is needed. Since a static_cast might still be needed.
This is the case for the ternary operator.
BUG: 407981
|
|
|
|
| |
CCBUG: 407623
|
| |
|
| |
|
|
|
|
| |
Differential Revision: https://phabricator.kde.org/D21178
|
| |
|
|
|
|
|
|
|
| |
Catches heap-allocating small trivially copyable/destructible types.
Such as: auto p = new QPoint();
Could contain false-positives, hence going to manual level.
|
| |
|
|
|
|
| |
So we can test generic QualTypes too
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
false-positives
This check has really been noise since its creation. The rate of false-positives
don't justify the insignificant compilation performance gains. Moved to manual level now.
Currently it has a bug, which doesn't seem possible to solve,
as the instantiated template doesn't carry over the information about the qint32 typedef.
And the CallExpr doesn't have any template related getters.
Also removed this check from the "performance" category in checks.json, since
this category is for runtime performance.
CCBUG: 407321
|
|
|
|
|
| |
It's commonly used to interact with Qt API.
This will be solved in Qt 6 anyway.
|