bug#9203: 24.0.50; some Clang experience

bug#9203: 24.0.50; some Clang experience

[Peter Dyballa]

Hello!

When using

	Apple clang version 2.0 (tags/Apple/clang-139) (based on LLVM 2.9svn)
	Target: x86_64-apple-darwin10
	Thread model: posix

to compile GNU Emacs on Mac OS X 10.6.8 with intel Core i7 it warns about some statements:

emacs-24.0.50/lib-src/../src/regex.c:6603:15: warning: 
      comparison of unsigned enum expression >= 0 is always true [-Wtautological-compare]
      if (ret >= 0)
          ~~~ ^  ~
emacs-24.0.50/lib-src/../src/regex.c:6619:14: warning: 
      comparison of unsigned enum expression >= 0 is always true [-Wtautological-compare]
  return ret >= 0 ? REG_NOERROR : REG_NOMATCH;
         ~~~ ^  ~

emacs-24.0.50/lib-src/emacsclient.c:1269:52: warning: 
      passing 'const char *' to parameter of type 'char *' discards qualifiers
                confstr (_CS_DARWIN_USER_TEMP_DIR, tmpdir, n);
                                                   ^~~~~~
In file included from emacs-24.0.50/lib-src/emacsclient.c:76:
In file included from ../lib/getopt.h:31:
In file included from /usr/include/getopt.h:44:
In file included from ../lib/unistd.h:43:
/usr/include/unistd.h:423:28: note: passing argument to parameter here
size_t   confstr(int, char *, size_t) __DARWIN_ALIAS(confstr);
                            ^

The installed NS variant is a bit unstable. Maybe I just need to optimise the compiler switches in the near future!

--
Greetings

  Pete

I hope to die before I *have* to use Microsoft Word.
			- Donald E. Knuth, 2001-10-02 in Tübingen

bug#9203: 24.0.50; some Clang experience

[Eli Zaretskii]

> From: Peter Dyballa <Peter_Dyballa@Freenet.DE>
> Date: Sat, 30 Jul 2011 13:53:30 +0200
> 
> When using
> 
> 	Apple clang version 2.0 (tags/Apple/clang-139) (based on LLVM 2.9svn)
> 	Target: x86_64-apple-darwin10
> 	Thread model: posix
> 
> to compile GNU Emacs on Mac OS X 10.6.8 with intel Core i7 it warns about some statements:
> 
> emacs-24.0.50/lib-src/../src/regex.c:6603:15: warning: 
>       comparison of unsigned enum expression >= 0 is always true [-Wtautological-compare]
>       if (ret >= 0)
>           ~~~ ^  ~

`ret' is of the type `reg_errcode_t', which is an enumerated data
type, so its signedness is implementation-defined, AFAIK.  Why did you
use that particular warning option?

bug#9203: 24.0.50; some Clang experience

[Peter Dyballa]

Am 30.07.2011 um 14:27 schrieb Eli Zaretskii:

>> From: Peter Dyballa <Peter_Dyballa@Freenet.DE>
>> Date: Sat, 30 Jul 2011 13:53:30 +0200
>> 
>> When using
>> 
>> 	Apple clang version 2.0 (tags/Apple/clang-139) (based on LLVM 2.9svn)
>> 	Target: x86_64-apple-darwin10
>> 	Thread model: posix
>> 
>> to compile GNU Emacs on Mac OS X 10.6.8 with intel Core i7 it warns about some statements:
>> 
>> emacs-24.0.50/lib-src/../src/regex.c:6603:15: warning: 
>>      comparison of unsigned enum expression >= 0 is always true [-Wtautological-compare]
>>      if (ret >= 0)
>>          ~~~ ^  ~
> 
> `ret' is of the type `reg_errcode_t', which is an enumerated data
> type, so its signedness is implementation-defined, AFAIK.  Why did you
> use that particular warning option?

I did not use any. Here is the configure invocation:

	env LANG=C PATH=/opt/local/bin:$PATH ./configure --without-sound --without-dbus --without-pop --without-gconf --without-gpm --with-ns --disable-ns-self-contained --enable-locallisppath=/Library/Application\ Support/Emacs/calendar24:/Library/Application\ Support/Emacs CFLAGS="-v -g -H -pipe -fPIC -fno-common -m64 -mtune=core2 -march=core2 -Os -fomit-frame-pointer -foptimize-register-move -ftree-vectorize" LDFLAGS="-Wl,-dead_strip_dylibs -Wl,-bind_at_load -Wl,-t" CC=clang CXX=clang++ PKG_CONFIG_PATH=/opt/local/lib/pkgconfig:/opt/local/share/pkgconfig:/usr/lib/pkgconfig

I thick Clang tells us how to avoid such reports. At least I don't get the one cited when I add -Wtautological-compare to the CFLAGS and configure and compile.

Purpose of my report is merely to point to possibly unclean code. Clangs really reports a lot!

--
Greetings

  Pete

There are three types of people in this world: those who can count, and those who cannot.

bug#9203: 24.0.50; some Clang experience

[Andreas Schwab]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Peter Dyballa <Peter_Dyballa@Freenet.DE>
>> Date: Sat, 30 Jul 2011 13:53:30 +0200
>> 
>> When using
>> 
>> 	Apple clang version 2.0 (tags/Apple/clang-139) (based on LLVM 2.9svn)
>> 	Target: x86_64-apple-darwin10
>> 	Thread model: posix
>> 
>> to compile GNU Emacs on Mac OS X 10.6.8 with intel Core i7 it warns about some statements:
>> 
>> emacs-24.0.50/lib-src/../src/regex.c:6603:15: warning: 
>>       comparison of unsigned enum expression >= 0 is always true [-Wtautological-compare]
>>       if (ret >= 0)
>>           ~~~ ^  ~
>
> `ret' is of the type `reg_errcode_t', which is an enumerated data
> type, so its signedness is implementation-defined, AFAIK.

But re_search returns regoff_t, so the type is wrong.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

bug#9203: 24.0.50; some Clang experience

[Eli Zaretskii]

> From: Andreas Schwab <schwab@linux-m68k.org>
> Cc: Peter Dyballa <Peter_Dyballa@Freenet.DE>,  9203@debbugs.gnu.org
> Date: Sat, 30 Jul 2011 15:22:11 +0200
> 
> > `ret' is of the type `reg_errcode_t', which is an enumerated data
> > type, so its signedness is implementation-defined, AFAIK.
> 
> But re_search returns regoff_t, so the type is wrong.

`ret' is the value returned by regexec, so its type should be
`reg_errcode_t'.

bug#9203: 24.0.50; some Clang experience

[Andreas Schwab]

Eli Zaretskii <eliz@gnu.org> writes:

> `ret' is the value returned by regexec

No, it isn't.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

bug#9203: 24.0.50; some Clang experience

[Eli Zaretskii]

> From: Andreas Schwab <schwab@linux-m68k.org>
> Cc: Peter_Dyballa@Freenet.DE,  9203@debbugs.gnu.org
> Date: Sat, 30 Jul 2011 15:49:32 +0200
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > `ret' is the value returned by regexec
> 
> No, it isn't.

Grow up, will you?

bug#9203: 24.0.50; some Clang experience

[Andreas Schwab]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Andreas Schwab <schwab@linux-m68k.org>
>> Cc: Peter_Dyballa@Freenet.DE,  9203@debbugs.gnu.org
>> Date: Sat, 30 Jul 2011 15:49:32 +0200
>> 
>> Eli Zaretskii <eliz@gnu.org> writes:
>> 
>> > `ret' is the value returned by regexec
>> 
>> No, it isn't.
>
> Grow up, will you?

Don't make you look silly.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

bug#9203: 24.0.50; some Clang experience

[Tom Tromey]

I looked at the warnings here and from what I can tell they've all been
fixed.

The regex.c one was fixed by correcting the type of 'ret'.

The emacsclient.c one was fixed by introducing tmpdir_storage and
passing it to confstr.

If you still see these warnings, please reply and we can reopen the bug.

thanks,
Tom