Re: "comparison always false" is a problem or not?

[Joel Rees]

On 2004.4.30, at 04:38 PM, Sherm Pendley wrote:

> On Apr 30, 2004, at 2:30 AM, Joel Rees wrote:
>
>> My experience is that this kind of thing tends to lead to dead code  
>> or endless loops. Do I need to dig in and find the macro declaration  
>> and see if I can fix it?
>>
>>     warning: comparison is always false due to limited range of data  
>> type
>
> In my own code, I compile with -Wall and try to chase down and  
> eliminate all warnings. Occasionally I might use a -Wno-something to  
> turn off a warning once I've determined that it's harmless, but I  
> really, really prefer to fix them.

Me too.

> I treat other people's code differently. If I'm building something  
> from source and it emits warnings, I'll make a mental note of that  
> fact. Then, if it's crashy, buggy, or exhibits some other odd  
> behavior, I might go back and see if the warnings are relevant to the  
> problems. If the app works fine though, I don't worry about the  
> warnings.

That's kind of what I was figuring would do, but I also would prefer  
not to find my code skipping a parse because the code got thrown out.  
So I thought I'd check.

> Those file names look familiar - are you building Perl?

Of course! (I suppose I should have said so.)

>  If so, and the self-tests pass,

Well, since you're kind enough to ask (heh heh), here's a few more I  
get:

In the make phase,

     ...
     cc -flat_namespace -L/usr/local/lib  -o miniperl ¥
         miniperlmain.o opmini.o libperl.a -lm -lc
     ./miniperl -w -Ilib -MExporter -e '<?>' || make minitest
     make: [extra.pods] Error 1 (ignored)
     ./miniperl -Ilib configpm configpm.tmp
     sh mv-if-diff configpm.tmp lib/Config.pm
     ...
         Making Errno (nonxs)
     Writing Makefile for Errno
     ../../miniperl "-I../../lib" "-I../../lib" "-I../../lib"  
"-I../../lib" Errno_pm.PL Errno.pm
     cp Errno.pm ../../lib/Errno.pm
     make: [extras.make] Error 1 (ignored)

             Everything is up to date. Type 'make test' to run test  
suite.

and also the range warnings mentioned already. In the make test phase,

     ...
     ext/DB_File/t/db-btree...............#
     # This test is known to crash in Mac OS X versions 10.2 (or earlier)
     # because of the buggy Berkeley DB version included with the OS.
     #
     FAILED at test 0
     ext/DB_File/t/db-hash................ok
     ext/DB_File/t/db-recno...............#
     # Some older versions of Berkeley DB version 1 will fail db-recno
     # tests 61, 63, 64 and 65.
     ...
     # You can safely ignore the errors if you're never going to use the
     ...
     #
     FAILED at test 64
     ext/Devel/DProf/t/DProf..............ok
     ...
     Failed 2 test scripts out of 804, 99.75% okay.
     ### Since not all tests were successful, you may want to run some of
     ...
     ### Bourne-style shells, like bash, ksh, and zsh, respectively.
     u=25.78  s=0  cu=599.71  cs=171.77  scripts=804  tests=80310
     make[2]: *** [_test_tty] Error 1
     make[1]: *** [_test] Error 2
     make: *** [test] Error 2

It says I can ignore those, so I did a make install and got

    ...
             Making Errno (nonxs)
     make[1]: [extras.make] Error 1 (ignored)

             Everything is up to date. Type 'make test' to run test  
suite.
     if [ -n "" ]; ¥
     then ¥
             cd utils; make compile; ¥
             cd ../x2p; make compile; ¥
             cd ../pod; make compile; ¥
     else :; ¥
     fi
     ./perl installperl --destdir=
     WARNING: You've never run 'make test' or some tests failed!  
(Installing anyway.)

which worried me a little. So I went ahead and ran t/harness after the  
install completed. (That's backwards, I suppose.):

     ...
     op/lex_assign..........................ok
     op/lfs.................................skipped
             all skipped: writing past 2GB failed: process limits?
     op/list................................ok
     ...
     ../ext/DB_File/t/db-btree..............#
     # This test is known to crash in Mac OS X versions 10.2 (or earlier)
     # because of the buggy Berkeley DB version included with the OS.
     #
     ../ext/DB_File/t/db-btree..............dubious
             Test returned status 0 (wstat 10, 0xa)
     ../ext/DB_File/t/db-hash...............ok
     ../ext/DB_File/t/db-recno..............#
     # Some older versions of Berkeley DB version 1 will fail db-recno
     # tests 61, 63, 64 and 65.
     #
     # For example Mac OS X 10.2 (or earlier) has such an old
     ...
     ../ext/DB_File/t/db-recno..............FAILED tests 64, 67, 70
             Failed 3/191 tests, 98.43% okay
     ../ext/Devel/DProf/t/DProf.............ok
     ...
     ../ext/Fcntl/t/syslfs..................skipped
             all skipped: writing past 2GB failed: process limits?
     ../ext/File/Glob/t/basic...............ok
     ...
     Failed Test                 Stat Wstat Total Fail  Failed  List of  
Failed
      
------------------------------------------------------------------------ 
-------
     ../ext/DB_File/t/db-btree.t    0    10    ??   ??       %  ??
     ../ext/DB_File/t/db-recno.t              191    3   1.57%  64 67 70
     59 tests and 486 subtests skipped.
     Failed 2/868 test scripts, 99.77% okay. 3/80321 subtests failed,  
100.00% okay.

Which I suppose looks pretty good, really, since the BDB stuff requires  
getting some good libraries, etc. But I did get some stuff skipped that  
I wanted to ask about, as well.

>  I definitely wouldn't worry about the warnings too much. Perl's  
> self-tests are remarkably thorough.

Anyway, I guess I was wondering how important those range warnings were  
likely to be.

Joel Rees