Ping?
On Wed, Sep 21, 2005 at 06:45:56PM -0700, Yitzchak Scott-Thoennes wrote:
> --- perl/embed.fnc.orig 2005-09-19 12:16:55.000000000 -0700
> +++ perl/embed.fnc 2005-09-21 13:05:46.762240000 -0700
> @@ -884,7 +884,7 @@
> p |int |yywarn |NN const char* s
> #if defined(MYMALLOC)
> Ap |void |dump_mstats |NN char* s
> -ApR |int |get_mstats |NN perl_mstats_t *buf|int buflen|int level
> +Ap |int |get_mstats |NN perl_mstats_t *buf|int buflen|int level
> #endif
> Anpa |Malloc_t|safesysmalloc |MEM_SIZE nbytes
> Anpa |Malloc_t|safesyscalloc |MEM_SIZE elements|MEM_SIZE size
> --- perl/cygwin/cygwin.c.orig 2005-07-08 10:03:01.000000000 -0700
> +++ perl/cygwin/cygwin.c 2005-09-21 13:25:50.022443200 -0700
> @@ -23,8 +23,8 @@
> Sigsave_t ihand,qhand;
> int childpid, result, status;
>
> - rsignal_save(SIGINT, SIG_IGN, &ihand);
> - rsignal_save(SIGQUIT, SIG_IGN, &qhand);
> + rsignal_save(SIGINT, (Sighandler_t) SIG_IGN, &ihand);
> + rsignal_save(SIGQUIT, (Sighandler_t) SIG_IGN, &qhand);
> childpid = spawnvp(_P_NOWAIT,path,argv);
> if (childpid < 0) {
> status = -1;
> End of Patch.
>
> Note that Perl_get_mstats always returns 0, so its return value isn't
> at all useful.
>
> This gets rid of:
>
> malloc.c:2447: warning: ignoring return value of `Perl_get_mstats', declared
> with attribute warn_unused_result
> cygwin.c:26: warning: passing arg 3 of `Perl_rsignal_save' from incompatible
> pointer type
> cygwin.c:27: warning: passing arg 3 of `Perl_rsignal_save' from incompatible
> pointer type
>
> leaving only:
>
> gv.c:167: warning: ignoring return value of `Perl_start_subparse', declared
> with attribute warn_unused_result
> sv.c:5568: warning: ignoring return value of `Perl_io_close', declared with
> attribute warn_unused_result
> perlio.c:648: warning: ignoring return value of `Perl_fp_dup', declared with
> attribute warn_unused_result
> byterun.c:893: warning: comparison is always false due to limited range of
> data type
>
> The gv.c warning I am unsure about; all other calls to start_subparse
> end up passing the return value to newATTRSUB, which passes it to
> LEAVE_SCOPE. I don't know if there is anything the code in gv.c
> (which is upgrading a prototype-only symbol table entry to be a full GV)
> should be doing with it.
>
> The sv.c and perlio.c warnings should probably just be suppressed.
> (How?) There isn't anything useful to do with an error return there.
>
> The byterun.c warning isn't worth trying to suppress IMO; maybe some
> day gcc will provide a nice way to suppress it. See:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12963
