h2xs generates incorrect code for Makefile.PL for enums #7602
Comments
From akolb@sun.comCreated by akolb@sun.comThe h2xs generates the following part of Makefile.PL for both #define and enum my $generate_code = This produce an entry in the Makefile.PL: my @names = (qw(bar_enum def_bar def_foo foo_enum)); ExtUtils::Constant::WriteConstants( The ExtUtils::Constant module treats all simple names as #defines and for #ifdef bar_enum for enums which do not have the name defined. Instead, for enum constants it my @names = (qw(def_bar def_foo), ExtUtils::Constant::WriteConstants( which will generate correct code for #defines and enums. Perl Info |
From sitz@onastick.netI got hit with this problem as well; the attached patch takes care of it |
From sitz@onastick.neth2xs.patch--- h2xs-cvs 2005-01-17 00:29:53.826930528 -0500
+++ h2xs 2005-01-17 00:40:06.392806360 -0500
@@ -859,7 +859,6 @@
# Remove C and C++ comments
$src =~ s#/\*[^*]*\*+([^/*][^*]*\*+)*/|("(\\.|[^"\\])*"|'(\\.|[^'\\])*'|.[^/"'\\]*)#$2#gs;
-
while ($src =~ /(\benum\s*([\w_]*)\s*\{\s([\s\w=,]+)\})/gsc) {
my ($enum_name, $enum_body) =
$1 =~ /enum\s*([\w_]*)\s*\{\s([\s\w=,]+)\}/gs;
@@ -869,8 +868,9 @@
for my $item (split /,/, $enum_body) {
my ($key, $declared_val) = $item =~ /(\w+)\s*(?:=\s*(.*))?/;
$val = defined($declared_val) && length($declared_val) ? $declared_val : 1 + $val;
+ print "Matched $key ($val)\n" if $opt_d;
$seen_define{$key} = $declared_val;
- $const_names{$key}++;
+ $const_names{$key} = {name => $key, macro => 1};
}
} # while (...)
} # if (!defined $opt_e or $opt_e)
@@ -1035,6 +1035,11 @@
}
}
my @const_names = sort keys %const_names;
+my @const_keys = map {
+ ref($const_names{$_}) eq 'HASH'
+ ? $const_names{$_}
+ : $_
+ } sort keys(%const_names);
-d $modpmdir || mkpath([$modpmdir], 0, 0775);
open(PM, ">$modpmname") || die "Can't create $ext$modpname/$modpmname: $!\n";
@@ -1422,7 +1427,7 @@
XS_FILE => $xsfallback,
DEFAULT_TYPE => $opt_t,
NAME => $module,
- NAMES => \@const_names,
+ NAMES => \@const_keys,
);
print XS "#include \"$constscfname\"\n";
}
@@ -1906,7 +1911,7 @@
XS_FILE => $constsxsfname,
DEFAULT_TYPE => $opt_t,
NAME => $module,
- NAMES => \@const_names,
+ NAMES => \@const_keys,
);
print PL <<"END";
if (eval {require ExtUtils::Constant; 1}) {
|
|
The RT System itself - Status changed from 'new' to 'open' |
From @rgsakolb@sun.com (via RT) wrote:
I suppose this is a duplicate of bug #32491, which has been fixed by |
From sitz@onastick.netOn Wed, Jan 19, 2005 at 12:30:51PM +0100, Rafael Garcia-Suarez wrote:
Am I missing the glaringly obvious, or should this: $seen_define{$key} = $declared_val; read like this: $seen_define{$key} = $val; ? --noah |
From @rgsNoah wrote:
You're not ; it should. Thanks, applied as change #23822 to bleadperl. |
From xw5s-73y5@disposable.spamcon.org
No it isn't. If you look at sitz's patch, you'll see that the version There were previously (at least) two distinct problems: 1) for some Please read Alexander's original description of the problem. sitz's |
From xw5s-73y5@disposable.spamcon.org
Never mind that part of my comment, I had misapplied the patch. |
|
@smpeters - Status changed from 'open' to 'resolved' |
Migrated from rt.perl.org#32498 (status was 'resolved')
Searchable as RT32498$
The text was updated successfully, but these errors were encountered: