Front page | perl.perl5.porters |
Postings from April 2000
Ilya Zakharevich <ilya@math.ohio-state.edu> wrote
> This is clearly a bug. Either \0 should not have been inserted, or
> (better) the length should have been marked as 10.
Since Z is defined to *always* insert a \0, the length should be 10.
> But this is a bug in pack(). I do not see why any change should be
> done to unpack().
Agreed. Patch attached.
Mike Guy
--- ./pp.c.orig Thu Apr 27 18:11:50 2000
+++ ./pp.c Thu Apr 27 18:04:17 2000
@@ -4445,7 +4445,8 @@
if ((*pat != 'a' && *pat != 'A' && *pat != 'Z') || pat[1] != '*')
DIE(aTHX_ "/ must be followed by a*, A* or Z*");
lengthcode = sv_2mortal(newSViv(sv_len(items > 0
- ? *MARK : &PL_sv_no)));
+ ? *MARK : &PL_sv_no)
+ + (*pat == 'Z' ? 1 : 0)));
}
switch(datumtype) {
default:
--- ./t/op/pack.t.orig Thu Apr 27 18:11:50 2000
+++ ./t/op/pack.t Thu Apr 27 18:12:05 2000
@@ -372,8 +372,9 @@
eval { ($x) = pack '/a*','hello' };
print 'not ' unless $@; print "ok $test\n"; $test++;
-$z = pack 'n/a* w/A*','string','etc';
-print 'not ' unless $z eq "\000\006string\003etc"; print "ok $test\n"; $test++;
+$z = pack 'n/a* N/Z* w/A*','string','hi there ','etc';
+print 'not ' unless $z eq "\000\006string\0\0\0\012hi there \000\003etc";
+print "ok $test\n"; $test++;
eval { ($x) = unpack 'a/a*/a*', '212ab345678901234567' };
print $@ eq '' && $x eq 'ab3456789012' ? "ok $test\n" : "#$x,$@\nnot ok $test\n";
End of patch
Thread Previous
|
Thread Next