Open lacasseio opened 5 years ago
Can you check-in this test and mark it @NotYetImplemented
or @Ignore
? We'll pick this up as part of our regular cycle
My suspicion is that we're not including the path to the source file as part of an implicit include path.
Maybe we talk about the same thing here. It seems to me that when we find an unresolved macro include, we give up on the analysis and potential relatively include headers. It would improve the fallback case. However, care should be taken to improve the analysis to avoid fallback scenario.
Expected Behavior
Relative included headers are part of the compile input header set fall back when encountering macro include.
Current Behavior
Header included relatively aren't considered as part of the fall back mechanic when encountering macro include.
Context
Given Gradle prone correctness, we should consider relative headers event when encountering unresolved macro includes.
Open Questions?
Should we only consider relative include from the discovered headers in the partial graph that is built or consider relative include from all headers in all include paths?
Steps to Reproduce (for bugs)
The following
CppIncrementalBuildIntegrationTest
test case show the issue:It fails at
relativeHeaderFile
change with the following error:Your Environment