At 10:17 PM +0000 1/18/08, Nicholas Clark wrote:
>On Thu, Jan 17, 2008 at 09:31:00AM -0600, Craig A. Berry wrote:
> > At 4:10 PM +0000 1/16/08, Nicholas Clark wrote:


> > >I'm particularly curious how it fares on VMS,
>>
>> @32985, here's what a default build gives me:
>>
> > t/io/fs.....................................FAILED at test 5


>$ diff -u t/io/fs.t ../../perl/t/io/fs.t
>--- t/io/fs.t   Sat Nov 17 20:18:43 2007
>+++ ../../perl/t/io/fs.t        Wed Sep 26 15:54:56 2007
>@@ -58,6 +58,7 @@
> }
> elsif ($^O eq 'VMS') {
>     `if f\$search("[.tmp]*.*") .nes. "" then delete/nolog/noconfirm 
> [.tmp]*.*.*`;
>+    `if f\$search("tmp.dir") .nes. "" then set file/prot=o:rwed tmp.dir;`;
>     `if f\$search("tmp.dir") .nes. "" then delete/nolog/noconfirm tmp.dir;`;
>     `create/directory [.tmp]`;
> }

That's harmless, but makes no difference to the test failure, which
has something to do with hard links:

not ok 5 - mode of triply-linked file
# Failed at [.io]fs.t line 114
#      got '488'
# expected '438'

I know I promised to look more closely at this, and I will, but haven't yet.

>That line is part of this, which I guess at the time I decided wasn't a
>good thing to integrate to maint. (Right or wrong?)

That's an excellent question.  I think we should take it because the
part that moves Perl_vms_start_glob from doio.c to vms.c is going to
be a prerequisite to a couple of other bug fixes to that routine.
There are at least three unrelated changes in that patch, which makes
it goshawful hard to figure out what should be integrated.  And this
raises the larger question of how much of the new VMS-specific code
should make its way back from blead to 5.8.x.  Nearly all of it
supports new VMS features, not new Perl features, so I guess there's
nothing that says we can't take it all.  FWIW, I do think we got
things pretty stable by the time 5.10 came out, and there's certainly
no prior patchlevel I can point to with confidence and say things
were better then.

>Change 27239 by [EMAIL PROTECTED] on 2006/02/20 09:31:03
>
>       Subject: [EMAIL PROTECTED] vms glob/readdir/chdir EFS/long filename 
> support
>       From: John E. Malmberg <[EMAIL PROTECTED]>
>       Date: Feb 20, 2006 3:43 AM
>       Message-ID: <[EMAIL PROTECTED]>
>
>Affected files ...
>
>... //depot/perl/doio.c#322 edit
>... //depot/perl/t/io/fs.t#59 edit
>... //depot/perl/vms/vms.c#162 edit
>
>Is the change to t/io/fs.t unrelated to the rest?


-- 
________________________________________
Craig A. Berry
mailto:[EMAIL PROTECTED]

"... getting out of a sonnet is much more
 difficult than getting in."
                 Brad Leithauser

Reply via email to