On Sun, 22 Nov 2015 00:19:15 +0100 Stefan Tauner stefan.tauner@alumni.tuwien.ac.at wrote:
Other workarounds:
- adding a dummy value to the enum with a negative value
- clang pragmas to disable the warnings
- a BUGS file
I am strictly against the pragmas because they litter the code too much. Carl-Daniel is against the -1 hack. I would not mind a BUGS file at all but I am also ok with just ignoring the issue. It does not happen in the default configuration and can easily be worked around by setting the CFLAGS (adding -Wno-tautological-compare), adding another programmer or using gcc instead...
Just a quick follow up... the previous patch was complete bullshit... the preprocessor does not know anything about enum values and replaced all those identifiers with 0 removing all #if-guarded lines - which "solves" the wrong warnings... because the lines are removed completely. I'll ignore the issue of clang overreacting in this case.