> > looks like it breaks lib/ExtUtils/t/MM_Unix.t (on Linux here)
> >
> > lib/ExtUtils/t/MM_Unix...............# Failed test
> (../lib/ExtUtils/t/MM_Unix.t at line 241)
> > # got: 'cc'
> > # expected: ''
> > FAILED at test 116
>
> Maybe this helps?
Yeah, sorry about that. In my rush to put out a patch I didnt think that test through at all. :-(
{LD} Should have a value after init_linker() no matter what. It could be either a presupplied value (not testable), the value in $Config{ld} or the default 'ld' (both obviously testable). So that test hypothetically could fail ($Config{ld} is set to '') even after the patch Jarkko provided.
Attched and below is a different remake of the test, that actually could be merged with Jarkkos anyway (although I didn't apply Jarkkos).
I really must get a cygwin perl development enviornment set up to cross test this stuff first.
Incidentally should the linker under linux be 'cc'? (Im assuming it should be as nobody raised that as a weirdness.)
My apologies for any frustration involved.
Yves
diff -u1 -wrd ExtUtils-MakeMaker-6.17\t/MM_Unix.t ExtUtils-MakeMaker-6.17-Patched-Test\t/MM_Unix.t
--- ExtUtils-MakeMaker-6.17\t/MM_Unix.t 2003-08-01 21:58:46.000000000 +0200
+++ ExtUtils-MakeMaker-6.17-Patched-Test\t/MM_Unix.t 2003-09-18 11:24:29.000000000 +0200
@@ -20,3 +20,3 @@
else {
- plan tests => 115;
+ plan tests => 116;
}
@@ -243,2 +243,4 @@
+use Config;
+ok(($t->{LD} eq $Config{ld})||($t->{LD} eq 'ld'),"ld was set to a known value");
diff -u1 -wrd ExtUtils-MakeMaker-6.17\t/MM_Unix.t
ExtUtils-MakeMaker-6.17-Patched-Test\t/MM_Unix.t
--- ExtUtils-MakeMaker-6.17\t/MM_Unix.t 2003-08-01 21:58:46.000000000 +0200
+++ ExtUtils-MakeMaker-6.17-Patched-Test\t/MM_Unix.t 2003-09-18 11:24:29.000000000
+0200
@@ -20,3 +20,3 @@
else {
- plan tests => 115;
+ plan tests => 116;
}
@@ -243,2 +243,4 @@
+use Config;
+ok(($t->{LD} eq $Config{ld})||($t->{LD} eq 'ld'),"ld was set to a known value");
