MacRuby warning: expression result unused

[llvmbot]

Bugzilla Link	3885
Resolution	WORKSFORME
Resolved on	Feb 22, 2010 12:41
Version	unspecified
OS	All
Reporter	LLVM Bugzilla Contributor

Extended Description

clang -I. -I./include -I/usr/include/libxml2 -arch i386 -arch x86_64 -fno-common -pipe -O2 -g -Wall -Wno-parentheses -Wno-deprecated-declarations -Werror -c iseq.c -o iseq.o
clang: warning: argument unused during compilation: '-fno-common'
clang: warning: argument unused during compilation: '-Wno-parentheses'
clang: warning: argument unused during compilation: '-Wno-deprecated-declarations'
iseq.c:345:45: warning: expression result unused
static inline VALUE CHECK_INTEGER(VALUE v) {NUM2LONG(v); return v;}
^~~~~~~~~~~
./include/ruby/ruby.h:384:34: note: instantiated from:
#define NUM2LONG(x) (FIXNUM_P(x)?FIX2LONG(x):rb_num2long((VALUE)x))
^
./include/ruby/ruby.h:227:21: note: instantiated from:
#define FIX2LONG(x) RSHIFT((SIGNED_VALUE)x,1)
^
./include/ruby/config.h:223:25: note: instantiated from:
#define RSHIFT(x,y) ((x)>>(int)y)
~~~^
1 diagnostic generated.
iseq.c:345:45: warning: expression result unused
static inline VALUE CHECK_INTEGER(VALUE v) {NUM2LONG(v); return v;}
^~~~~~~~~~~
./include/ruby/ruby.h:384:34: note: instantiated from:
#define NUM2LONG(x) (FIXNUM_P(x)?FIX2LONG(x):rb_num2long((VALUE)x))
^
./include/ruby/ruby.h:227:21: note: instantiated from:
#define FIX2LONG(x) RSHIFT((SIGNED_VALUE)x,1)
^
./include/ruby/config.h:223:25: note: instantiated from:
#define RSHIFT(x,y) ((x)>>(int)y)
~~~^
1 diagnostic generated.

[lattner]

Can you include a .i file please? Thanks.

[llvmbot]

iseq.i

[lattner]

Ok, expanded out, this is:

static inline VALUE CHECK_INTEGER(VALUE v) {((((long)(v))&RUBY_FIXNUM_FLAG)?(((long)v)>>(int)1):rb_num2long((VALUE)v)); return v;}
~~~~~~~~~^ ~~~~~~

I think that the warning is correct, do you agree?

[lattner]

In other words, when the condition is true, the shift is evaluated, but the result (passed to the ?:) is ignored.

[llvmbot]

In other words, when the condition is true, the shift is evaluated, but the
result (passed to the ?:) is ignored.

I didn't complain about the warning as much as I think saved-temp version
is much more readable. I guess you probably can't print the saved-temp version of
the warning and followed by all the MACROS like it is doing right now.

[lattner]

Right. Hey, at least we do better than GCC, which probably prints the moral equivalent of:

iseq.c:345: warning: expression result unused
static inline VALUE CHECK_INTEGER(VALUE v) {NUM2LONG(v); return v;}

:)

-Chris