>>locale there is some test noise that doesn't seem to affect success.
>
> I can't recreate this. I've tried with both nl_NL.utf8 and fa_IR.utf8
> with no joy
>
>>The curious part is that if the test is run manually, no noise:
>>
>>.../t ; env PERL_CORE=1 LC_ALL=fi_FI.UTF-8 ./perl -I../lib ../ext/B/t/xref.t
>
> You need a -C flag in there, don't you?
Ooops. Who idiot ever thought up such a stupid interface? Oh, right,
never mind :-) Yes, with -C I get it:
1..14
ok 1 - use B::Xref;
ok 2 - compile() returns a coderef (CODE(0x1402f4d90))
utf8 "\xCB" does not map to Unicode at ../ext/B/t/xref.t line 205,
<XREF> line 1045.
utf8 "\xB8" does not map to Unicode at ../ext/B/t/xref.t line 205,
<XREF> line 1104.
ok 3 - $compilesub present in main program
...
Sadly, no valgrind for Tru64. There's the "third" tool, but it
doesn't show any funny business going on with the xref.t.
I managed to cut down the xref.t (I lost one of the two warnings,
though, hopefully one is enough) to the minimal version that still
gives the "\xB8" warning (with env LC_ALL=fi_FI.UTF-8 ./perl -C -Ilib):
use warnings;
use B::Xref;
open XREF, 'xreftest.out';
while (<XREF>) {
}
The offending line of xreftest.out is (through Emacs' hexl-mode):
00008270: 2020 2020 2020 2020 2020 2020 2020 3634 64
00008280: 0a46 696c 6520 b80a 2020 5375 6272 6f75 .File .. Subrou
00008290: 7469 6e65 2028 6465 6669 6e69 7469 6f6e tine (definition
The "File \xb8\n" is the badness. No I don't know where "File \xb8"
might be coming from. I know (or have a good guess of) which file
it *was*, if we look at the data following it:
File ?
Subroutine (definitions)
Package overload
&AddrRef s94
&Method s82
&OVERLOAD s29
&Overloaded s55
&OverloadedStringify s73
&StrVal s100
&constant s160
&import s36
&mycan s69
&nil s7
&ov_method s63
&remove_constant s169
&unimport s49
It was lib/overload.pm, I would guess...
--
Jarkko Hietaniemi <[EMAIL PROTECTED]> http://www.iki.fi/jhi/ "There is this
special
biologist word we use for 'stable'. It is 'dead'." -- Jack Cohen
