Created attachment 108658 [details] Sementic Analysis of messagepart.cpp I try to scan kde/pim/messagelib for the semantic analysis - in detail mimetreeparser/src/messagepart.cpp compiles correctly with gcc/make, but the semantic analysis, tells me that MessagePart is incomplete in messagepart.cpp. But messagepart.h is as first imported with the complete class. The later import of objecttreeparser.h has correctly only a forward class declaration.
More things look off here, apparently it doesn't find the Qt includes. Did you open the project with the CMake project manager ...?
I used project->Open/Import Project and selected the CMakeLists.txt. Maybe also a good information: other files in the same project are correctly analyzed. The header file messagepart.h looks corrercly analized from the colors, but it complains about missing mimetreeparser/util.h, mimetreeparser/bodypartformatter.h, mimetreeparser_export.h. The build directory is correct - this is the one I build the software successfully. I use a neon docker container for kdevelop and building. a funny side thing, in messagepart.cpp the include of mimetreeparser_debug.h is successfull but at an other file (nodehelper.cpp) in the same directory the analyzer complains about missing mimetreeparser_debug.h. the CMakeLists.txt part: set(libmimetreeparser_SRCS messagepart.cpp nodehelper.cpp [...] ) add_library(KF5MimeTreeParser ${libmimetreeparser_SRCS} ) generate_export_header(KF5MimeTreeParser BASE_NAME mimetreeparser) add_library(KF5::MimeTreeParser ALIAS KF5MimeTreeParser) set(mimetreeparser_LINK_LIBRARIES ) target_link_libraries(KF5MimeTreeParser PRIVATE QGpgme KF5::Codecs KF5::I18n KF5::CoreAddons KF5::Mime Qt5::Gui )
Created attachment 108664 [details] only some files are analyzed In the screeshot you see an other day opening kdevelop. Interessting is that inside MessagePartPrivate the analyzier gets better results, but still includes are not handled correcly. But after the closing the namespace the analyzier stops completly.
With 5.2.0 the issue is fixed.
Nice, thanks for the report. Probably was fixed by the work on adding the correct compiler flags from cmake.