Hi,

> > As I already wanted to have a look at Perl 5.6, I installed the
> > CPAN version yesterday. It went fine exceptfor six tests - I'll look
> > into these over the weekend. Environment is DEC C 5.7-004 and MMK
> > on VMS/VAX 6.2, configuring the standard options (except no -D).
>
> I too saw several `MMS test` failures.  I am pretty sure that
> proposed patches for most if not all the failures have been posted
> to the vmsperl list.  A few did not make it into 5.6.0 unfortunately.

Yup. Unfortunately, I'm still in catch-up mode with the list. So I looked
into failures that already have been patched ([.lib]glob-basic.t).

With [.lib]vmsish.t, I have the same effect as John Peacock (see
<[EMAIL PROTECTED]>. When run as part of 'mmk test',
it reports 'FAILED on test 7', while when run directly, it passes test
7, but barks on

not ok 17  # (stat) UTC: 955127328  VMS: 955134528

The times look OK, IMHO (I have MET DST, so SYS$TIMEZONE_DIFFERENTIAL=7200).
I'm not sure of the formula, though:

  $vmsmtime - $utcmtime + $offset

that is, $vmsmtime should be the same as ($utcmtime - $offset) ?!
I think it should be ($utcmtime + $offset), shouldn't it? Remember, east
of Greenwich counts as positive.

What puzzles me is that all other tests pass - although they apply
the same formula...

Other things that went wrong were [.op]read.t (got wrong record format when
re-packing the distribution as ZIP. Fixed by CONVERTing it to Stream_LF),
[.pod]pod2usage.t and podselect.t ('FAILED on test 1' and hang(!) with
'mms test', OK when run directly), and [.pragma]warnings ('FAILED on test
216'
with 'mms test', OK when run directly).

> > XML::Parser::Expat's DESCRIP.MMS still looks the same, i.e. the bug
> > is still there. <sigh>
>
> That is too bad.  Do you think a modification to MM_VMS.pm is in order,
> or could it be worked around with changes to the Makefile.PL?

I'll have yet to look deeper into things, e.g. why the author thinks he
needs that C and OBJECT attributes, and why MM_VMS munges things up.
If I only had the time I'd like to have...

cu,
  Martin
--
 OpenVMS:         | Martin Vorlaender         VMS & WNT programmer
 When you KNOW    | work: [EMAIL PROTECTED]
 where you want   |       http://www.pdv-systeme.de/users/martinv/
 to go today.     | home: [EMAIL PROTECTED]

Reply via email to