GCC 7 added a new diagnostic, -Wimplicit-fallthrough, which is enabled
with -Wextra that warns about implicitly falling through a case
statement.
[ 4%] Building C object CMakeFiles/msgpackc-static.dir/src/unpack.c.o
/usr/lib/gcc-snapshot/bin/gcc -I/home/jamessan/src/msgpack-c/. -I/home/jamessan/src/msgpack-c/include -I/home/jamessan/src/msgpack-c/build/include -g -O2 -fdebug-prefix-map=/home/jamessan/src/msgpack-c=. -specs=/usr/share/dpkg/no-pie-compile.specs -fstack-protector-strong -Wformat -Werror=format-security -Wdate-time -D_FORTIFY_SOURCE=2 -Wall -Wextra -Werror -g -O3 -o CMakeFiles/msgpackc-static.dir/src/unpack.c.o -c /home/jamessan/src/msgpack-c/src/unpack.c
In file included from /home/jamessan/src/msgpack-c/src/unpack.c:283:0:
/home/jamessan/src/msgpack-c/include/msgpack/unpack_template.h: In function 'template_execute':
/home/jamessan/src/msgpack-c/include/msgpack/unpack_template.h:238:17: error: this statement may fall through [-Werror=implicit-fallthrough=]
++p;
^~~
/home/jamessan/src/msgpack-c/include/msgpack/unpack_template.h:240:13: note: here
default:
^~~~~~~
cc1: all warnings being treated as errors
Adding the comment makes it explicit that the fallthrough is
intentional, so gcc doesn't complain.
For EXT 8/16/32, the "size" field was being incremented by 1 to account for the
type field, but according to the specification the size should only consider the
length of the data field.
Petr Benes