* Fix#11223 checkLibraryFunction treats "auto" as type
* Use utility function
* Get type from 'auto p = new ...'
* Fix merge
* Add test for recent fix
* Format
* Get type from smartpointer
* Make control flow a bit easier, and more similar to previous code
Made similar to around line 790
* In a cpp11init, always parse only the corresponding } (#11134)
- _always_, because in some cases this was omitted (around line 790) or too strict (around line 860)
- _only_, and not following tokens which happen to be } as well (around line 1030)
* Fix unit tests: AST was incorrect, now is fixed
auto var{ {{},{}}, {} };
Old AST:
```
{
|-var
`-{
`-,
|-,
| |-{
| `-{
`-{
```
New AST:
```
{
|-var
`-,
|-{
| `-,
| | |-{
| | `-{
`-{
```
Compare the same example, but with `X{}` instead of just `{}`:
`auto var{ a{b{},c{}}, d{} };`
```
{
|-var
`-,
|-{
| |-a
| `-,
| | |-{
| | | `-b
| | `-{
| | | `-c
`-{
`-d
```
This structure is similar to that of the new AST, not the old AST
* Fix unit tests: another AST was incorrect, now is fixed
Code: `auto var{{1,a::b{2,3}}, {4,a::b{5,6}}};`
Old AST:
```
{
|-var
`-{
`-,
|-,
| |-1 'signed int'
| `-{
| | |-::
| | | |-a
| | | `-b
| | `-,
| | | |-2 'signed int'
| | | `-3 'signed int'
`-{
`-,
|-4 'signed int'
`-{
|-::
| |-a
| `-b
`-,
|-5 'signed int'
`-6 'signed int'
```
New AST:
```
{
|-var
`-,
|-{
| `-,
| | |-1 'signed int'
| | `-{
| | | |-::
| | | | |-a
| | | | `-b
| | | `-,
| | | | |-2 'signed int'
| | | | `-3 'signed int'
`-{
`-,
|-4 'signed int'
`-{
|-::
| |-a
| `-b
`-,
|-5 'signed int'
`-6 'signed int'
```
* Fix unit tests: missing ; after class, resulting in incorrectly being marked as cpp11init
Because of the missing `;` after the class declaration, it was marked as a cpp11init block.
Which it isn't, and which now throws an exception
* Fix cpp11init to let unit tests pass again
The following unit tests failed on the newly introduced throws, because the code for these tests incorrectly marked some tokens as cpp11init:
TestVarID::varid_cpp11initialization
TestTokenizer::checkRefQualifiers
* Fix typo
* Improve check for void trailing return type
Observation: the only function body _not_ containing a semicolon, is a void function: something like
auto make_zero(ini& i) -> void {
while(--i > 0) {}
}
Non-void function? Then it must return a value, and thus contain a semicolon, which is checked for a few lines later.
* Fix cpp11init with templated trailing return type
In the following example, vector was marked as cpp11init due to the mismatch of `%any% {`
auto f() -> std::vector<int> { return {}; }
I made the assumption that whenever "%any% {" matches, endtok must be set too.
If this assumtion doesn't hold (so "%any% {" matches, but endtok == nullptr), then the for-loop would search all the way to the end of stream. Which I guess was not the intention.
* Remove comments
Co-authored-by: Gerbo Engels <gerbo.engels@ortec-finance.com>
In this example:
```
//template<std::same_as<int> T> // <= works
template<same_as<int> T> // <= didn't work
void f()
{}
```
the changed line used to match to `< same_as <`, therefore skip creating links.
The `%op% %name% <` already feels a bit like a workaround. So adding the condition that $op$ shouldn't be a comparison operator, but part of the template, seemed reasonable to me
Co-authored-by: Gerbo Engels <gerbo.engels@ortec-finance.com>
* Fix 9392, but for destructors: out-of-line defaulted destructors skipped everything after
Context:
```
struct S {
~S();
};
S::~S() = default;
void g() {
int j;
++j;
}
```
Everything after `S::~S() = default;` was skipped, so the uninitialized variables in g() weren't found.
Out-of-line destructors are useful e.g. when you have a forward declared unique_ptr in the .h,
and `= default` the destructor in the .cpp, so only the cpp needs to know the header for destructing
your unique_ptr (like in the pImpl-idiom)
* Fix unit test, by correctly fixing 10789
Previous commit broke this test, but also provided the tools for a cleaner fix
* Document current behaviour
* Rewrite control flow
* Fix deleted functions, which skipped everything after
`a::b f() = delete` triggered the final else in SymbolDatabase::addNewFunction,
which sets tok to nullptr, effectively skipping to the end of the stream.
* Remove troublesome nullptr, which skips every analysis afterwards
It was introduced in 0746c241 to fix a memory leak.
But setting tok to nullptr, effectively skipping to the end, seems not needed.
Previous commits fixes prevented some cases where you could enter the `else`.
This commit is more of a fall back.
* fixup! Fix deleted functions, which skipped everything after
`a::b f() = delete` triggered the final else in SymbolDatabase::addNewFunction,
which sets tok to nullptr, effectively skipping to the end of the stream.
* fixup! Fix deleted functions, which skipped everything after
`a::b f() = delete` triggered the final else in SymbolDatabase::addNewFunction,
which sets tok to nullptr, effectively skipping to the end of the stream.
* Make it heard when encountering unexpected syntax/tokens
Co-authored-by: Gerbo Engels <gerbo.engels@ortec-finance.com>
* Fix internalAstError with new
* Format
* nullptr check
* Add test for #11039
* Fix#11039 Empty AST with delete new / #11327 FP leakReturnValNotUsed with new and offset
* cleaned up triage.pro and pass all triage files to selfcheck
* CI-unixish.yml: factored out redundant selfcheck flags
* CI-unixish.yml: cleaned up selfcheck suppressions and fixed warnings
* added explicit `missingInclude` checks to selfcheck to work around current issues/limitations / fixed selfcheck includes / added some unit tests for `missingInclude`
* applied the selfcheck changes to the sanitizer jobs
* only fail TSAN CI build in case of TSAN error / updated sanitizer CI build TODOs
chrchr-github
Stefan van Kessel
gerboengels
Daniel Marjamäki
Oliver Stöneberg
Paul Fultz II