Thou shalt have no other gods before the ANSI C standard 1605

Jean-Marc Bourguet

Usually I use source-line usage counters, as in the following Makefile target which exploits "ctrace":

$(PCFILES): $(HFILES) $(CFILES) $(OBJS) $(TEST).o # $(CFILES) is overkill o=`basename $$c .c`.o; $(CC) -c T.$$c; mv $$o B.$$o && mv T.$$o $$o; $(CC) $(LDFLAGS) -o $(TEST) $(TEST).o $(OBJS) $(TLIBES); rm -f $(TEST) T.$$o && mv B.$$o $$o; sed 's-^ *-(1-90-9*-).*--1-' rm -f X.$$c; ( echo ',x-tct-("----n *1-90-9*{'; echo 'x-1-90-9*-p-n---n-p'; echo '}-nq-n' ( echo f P.$$c; echo ',x-^-c-'; uniq -c L.$$c sed 's;^-( *1-90-9*-)(1-90-9*-);-2i--1-;'; uniq L.$$c comm -13 - A.$$c sed 's;.*;&i 0-;'; echo 'w-nq-n' ) sam -d $$c; rm -f ALT.$$c )

I'd split it up into if ( a b ) g1(); which should show separate counters for both conditions.

I look for the number of ends of the controlled expression (g1() or g2()) being less than the number of ends of the controlling condition (a). Another method is to always code explicitly the other branch: if ( a ) g1(); else ; etc. Anyway, the coverage tests are devised by looking at the source code, so it is rare that usage counters show that some branch has not been exercised.