This is part of the implementation of the dataflow analysis framework.
See "[RFC] A dataflow analysis framework for Clang AST" on cfe-dev.
Details
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
clang/lib/Analysis/FlowSensitive/Transfer.cpp | ||
---|---|---|
115 | An alternative way to handle Noop operations would be to make location lookup function always skip certain nodes, so we do not need to store locations for those subexpressions. I don't have a strong feeling for either solution, this is fine as it is, just wanted to be sure that both were considered. | |
287 | Interesting, I only see isCallToStdMove in the CallExpr API, although I imagine, std::forward has a similar level of importance. | |
clang/unittests/Analysis/FlowSensitive/TransferTest.cpp | ||
40 | Shouldn't we piggyback on clang::LangStandard::Kind? If it is not easy to convert that to the appropriate command line flag feel free to ignore this. |
clang/lib/Analysis/FlowSensitive/Transfer.cpp | ||
---|---|---|
115 | It makes sense to consider it. Right now I don't have signals that one is better than the other. I think the current implementation fits the general model a bit better. Nevertheless, I added a FIXME to keep this option in mind. | |
287 | I haven't looked into std::forward yet, but happy to do that in the next patch. | |
clang/unittests/Analysis/FlowSensitive/TransferTest.cpp | ||
40 | Makes sense. Thanks for pointing this out! |
clang/unittests/Analysis/FlowSensitive/TransferTest.cpp | ||
---|---|---|
1362 | I think this changed from 14 to 17 in the last revision. Is this intentional? |
Address reviewers' comments.
clang/unittests/Analysis/FlowSensitive/TransferTest.cpp | ||
---|---|---|
1362 | Not intentional. Thanks for catching it! |
clang/lib/Analysis/FlowSensitive/Transfer.cpp | ||
---|---|---|
99–106 | use a switch? |
clang/lib/Analysis/FlowSensitive/Transfer.cpp | ||
---|---|---|
99–106 | Refactored to use a switch. |
use a switch?