| Summary: | C++11 lambdas: visibility/capture-lists not handled correctly | ||
|---|---|---|---|
| Product: | [Applications] kdevelop | Reporter: | Jörg Pfähler <bluecoder> |
| Component: | Language Support: CPP (old) | Assignee: | kdevelop-bugs-null |
| Status: | RESOLVED FIXED | ||
| Severity: | minor | CC: | aleixpol |
| Priority: | NOR | ||
| Version: | git master | ||
| Target Milestone: | --- | ||
| Platform: | Unlisted Binaries | ||
| OS: | Linux | ||
| Latest Commit: | Version Fixed In: | ||
| Sentry Crash Report: | |||
yep, needs proper DUChain support. thanks for the report Git commit 05b647a4d555c2e406a08b6b9babacdc4b6ef105 by Milian Wolff. Committed on 18/01/2012 at 22:17. Pushed by mwolff into branch '4.3'. c++11 support: first work towards proper DUChain integration of lambda expressions - a lambda expression now properly gets a function type associated - the lambda declarator now is represented by a function context which gets imported into the body of the lambda expression known problems: - the surrounding context is still implicitly imported into the body context - the return-type is not properly computed in cases where no explicit trailing return type is given but only a single return statement in the body instead - lambda captures are still not handled at all M +16 -0 languages/cpp/cppduchain/contextbuilder.cpp M +1 -0 languages/cpp/cppduchain/contextbuilder.h M +23 -0 languages/cpp/cppduchain/expressionvisitor.cpp M +1 -0 languages/cpp/cppduchain/expressionvisitor.h M +1 -0 languages/cpp/cppduchain/tests/test_duchain.h M +45 -0 languages/cpp/cppduchain/tests/test_duchain_cpp2011.cpp http://commits.kde.org/kdevelop/05b647a4d555c2e406a08b6b9babacdc4b6ef105 Git commit 40bd4b6057be525905a14f953900dc1b53eef900 by Milian Wolff. Committed on 18/01/2012 at 23:09. Pushed by mwolff into branch '4.3'. c++11 support: properly report uses for lamba captures M +17 -0 languages/cpp/cppduchain/expressionvisitor.cpp M +1 -0 languages/cpp/cppduchain/expressionvisitor.h M +1 -0 languages/cpp/cppduchain/tests/test_duchain.h M +37 -1 languages/cpp/cppduchain/tests/test_duchain_cpp2011.cpp M +1 -1 languages/cpp/parser/ast.h M +4 -2 languages/cpp/parser/parser.cpp http://commits.kde.org/kdevelop/40bd4b6057be525905a14f953900dc1b53eef900 proper visibility/capture support looks tough, lets see when I can manage this. the current situation should be better already though I hope. Hello! We are working on a new clang-based C/C++ language plugin for KDevelop 5 which supersedes the old C++ plugin in KDevelop 4. See e.g.: https://www.kdevelop.org/news/first-beta-release-kdevelop-500-available Due to a lack of manpower, we cannot fix bugs in the old C++ plugin. We rather want to supply a good Clang based C++ experience for KDevelop 5 than wasting our time on the legacy C++ support for KDevelop 4. With the new clang-based C/C++ language plugin, the bug presented here does not occur. In my testing. For these reasons, I'll close this bug. Please stay tuned for KDevelop 5. If you think this bug is applicable to Clang/KDevelop 5, please reopen the report and add new information on how to reproduce the bug there. |
Version: git master (using KDE 4.7.0) OS: Linux Consider the following code: int i; std::function<void (int)> f = [] (int i){i = 0;}; kdevelop now highlights the i outside of the lambda at the same time as the one inside the lambda (but not that one in the argument list). This is clearly wrong (or at least misleading) since the i in the argument list is the same as the one in the body, but not the same as the global one. Another related bug/feature request is the following: int i; std::function<void ()> f = [&i] (){i = 0;}; I would except now, that the i in the capture list is highlighted the same way as the i in outside and inside the lambda. Reproducible: Always Steps to Reproduce: - Expected Results: -