64-bit work. Now 32-bit platforms get a 100% make test

with -Duse64bits (using long long).
Tested in Solaris 2.6 sparc  RH Linux 6.0 x86
(and Digital IX 4.0D, to get a true 64-bit opinion).  Now e.g.
'print unpack "q", pack "q", 12345678901'
should work on such 32-bit platforms.
Still a lot of printf()s behind -D which wrongly assume
that %ld/%lx and (long) are a good combination.
Introducing a slew of new macros intended to be used in printf()
format strings: e. g. PERL_PRId64 is the string to be used
when printing an IV, printf("%" PERL_PRId64 "\n", iv).
The PRI... naming follows the C9X naming of <inttypes.h> macros.

p4raw-id: //depot/cfgperl@3861

1 files changed, 6 insertions, 6 deletions

diff --git a/t/pragma/warn/sv b/t/pragma/warn/sv
index 0421192104..b6c91c9fe9 100644
--- a/t/pragma/warn/sv
+++ b/t/pragma/warn/sv
@@ -235,24 +235,24 @@ Subroutine fred redefined at - line 5.
 # sv.c
 use warning 'printf' ;
 open F, ">".($^O eq 'VMS'? 'NL:' : '/dev/null') ;
-printf F "%q\n" ;
-my $a = sprintf "%q" ;
+printf F "%z\n" ;
+my $a = sprintf "%z" ;
 printf F "%" ;
 $a = sprintf "%" ;
 printf F "%\x02" ;
 $a = sprintf "%\x02" ;
 no warning 'printf' ;
-printf F "%q\n" ;
-$a = sprintf "%q" ;
+printf F "%z\n" ;
+$a = sprintf "%z" ;
 printf F "%" ;
 $a = sprintf "%" ;
 printf F "%\x02" ;
 $a = sprintf "%\x02" ;
 EXPECT
-Invalid conversion in sprintf: "%q" at - line 5.
+Invalid conversion in sprintf: "%z" at - line 5.
 Invalid conversion in sprintf: end of string at - line 7.
 Invalid conversion in sprintf: "%\002" at - line 9.
-Invalid conversion in printf: "%q" at - line 4.
+Invalid conversion in printf: "%z" at - line 4.
 Invalid conversion in printf: end of string at - line 6.
 Invalid conversion in printf: "%\002" at - line 8.
 ########