Nicholas Clark wrote:
> On Tue, Nov 18, 2008 at 09:37:50AM -0000, Steve Hay wrote:
>> The deparse.t failures were:
>>
>> not ok 63 - tests for deparsing imported constants
>> # Failed test 'tests for deparsing imported constants'
>> # at ../ext/B/t/deparse.t line 94.
>> # '{
>> # warn &O_NONBLOCK();
>> # }'
>> # doesn't match '(?-xism:^\{\s*\s+warn\s+O_NONBLOCK\;\s*\}$)'
>>
>> not ok 65 - tests for deparsing imported constants that got deleted
>> from the original namespace # Failed test 'tests for deparsing
>> imported constants that got deleted from the original namespace' #
>> at ../ext/B/t/deparse.t line 94. # '{
>> # warn &O_SYNC();
>> # }'
>> # doesn't match '(?-xism:^\{\s*\s+warn\s+O_SYNC\;\s*\}$)'
>
> On Win32 both POSIX::O_NONBLOCK and POSIX::O_SYNC are actually
> subroutines that croak() with a "not defined" if you call them?
Yes, I think so:
C:\p5p\bleadperl>.\perl -MPOSIX -e "POSIX::O_NONBLOCK()"
Your vendor has not defined Fcntl macro O_NONBLOCK, used at -e line 1.
C:\p5p\bleadperl>.\perl -MPOSIX -e "POSIX::O_SYNC()"
O_SYNC is not a valid POSIX macro at -e line 1
> If "yes" - can you suggest two Fcntl constants re-exported by POSIX> that are defined on Win32?These are all okay: O_APPEND, O_CREAT, O_EXCL, O_TRUNC.
Highlight