develooper Front page | perl.vmsperl | Postings from June 2002

Re: Not OK: perl v5.8.0 +DEVEL17237 on VMS_AXP V7.2-1

Craig A. Berry
June 14, 2002 16:24
Re: Not OK: perl v5.8.0 +DEVEL17237 on VMS_AXP V7.2-1
Message ID:
At 3:05 PM -0400 6/14/02, John Peacock wrote:
>This is a build failure report for perl from jpeacock@kes.internal,
>generated with the help of perlbug 1.33 running under perl v5.8.0.
>lib/ExtUtils/t/Command...............FAILED at test 10
>lib/File/Spec/t/rel2abs2rel..........FAILED at test 4
>lib/File/stat........................FAILED at test 18

Hmm.  I seem to remember there were more last time you posted, though
I can't remember any of this week's patches that would have fixed
anything in your configuration.  Oh well, good news is good news.

> >perl [.lib.File.Spec.t]rel2abs2rel.t
>Can't pipe "[-.user.jpeacock.perlsrc.perl]perl.exe;1": no
>such file or directory at [.lib.file.spec.t]rel2abs2rel.t line 43.
>not ok 4 - canonpath on abs executable
>#     Failed test ([.lib.file.spec.t]rel2abs2rel.t at line 65)
>#          got: ''
>#     expected: 'ok
># '

> The middle error only occurs when running from a
>rooted logical.  Where is the code which is used to set $^X?  The problem
>is that the implied CWD is rooted and $^X is not, so the abs2rel code gets

I think it comes straight out of argv[0].  We could break it into
pieces using sys$filescan and put the pieces back together using the
current default device (which might be a rooted logical), but that
would be an unsightly hack that could easily cause more breakage than
it solves.  I'm not sure how best to handle this one.

Craig A. Berry

"... getting out of a sonnet is much more
 difficult than getting in."
                 Brad Leithauser Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About