Front page | perl.vmsperl |
Postings from June 2002
RE: possible VMS v5.5 fix (was RE: 5.8.0 release candidate status)
From:
Bohan, Larry
Date:
June 13, 2002 13:41
Subject:
RE: possible VMS v5.5 fix (was RE: 5.8.0 release candidate status)
Message ID:
EA2A66B6B16B6F44AD8F53E6D80000A803E3F7@srvrxchange
>t/io/fs..............................FAILED at test 31
i recall making a trivial/small chg to [.t.op]fs.t
to work-around this. but I don't have the original version at hand,
would have go unpack the orig tarball to compare,
and get a real gnu-diff back. but the chg was roughly
like so, wrapped in a if-VMS ...
# check if rename() works on directories
if ($^O eq 'VMS') {
# must have delete access to rename a directory
`set file tmp.dir/protection=o:d`;
ok(rename('tmp.dir', 'tmp1.dir'), "rename on directories1") ||
print "# errno: $!\n";
} else {
ok(rename('tmp', 'tmp1'), "rename on directories2");
}
also, ISTR some instances of 'tmp.dir;2' being created.
but I'm sorry to say, didn't poke around long enough to
puzzle out which test had created them.
>t/op/stat............................FAILED at test 29
this one, iirc, I saw when the [perl...] tree
was not owned by the user/account I was doing the make/test/install from.
I believe it was a "ok(-o $tmpfile,'-o');" (effective ownership?)
test in [t.opt]stat.t
When RC2 hits the streets, I'll comb though thru these 2 tests
better detail. My interest in RC1 had primarily been a 1st pass,
see how easy it was to rebuild various extensions (30-45 of 'em),
and take notes on what work-arounds were needed.
-----Original Message-----
From: Craig A. Berry [mailto:craigberry@mac.com]
Sent: Thursday, June 13, 2002 1:10 PM
To: Henry A. Frystak - System Manager
Cc: vmsperl@perl.org
Subject: RE: possible VMS v5.5 fix (was RE: 5.8.0 release candidate
status)
At 1:35 PM -0400 6/13/02, Henry A. Frystak - System Manager wrote:
> Thanks for the quick response. I've been a busy beaver, but I'm
>about to run out of TUIT credits. The good news and the bad news: Your
patch
>applied cleanly and got past the original problem.
Good, thanks for the confirmation.
> However, I ran into another
>problem further into the make cycle. The appropriate lines are included
below:
>CC/DECC /Include=[]/Standard=Relaxed_ANSI/Prefix=All/Obj=.obj
/Define=("VERSION=""0.01""","XS_VERSION=""0.01""")/Include=([---])
> /NoList LANGINFO.c
> if ((s = nl_langinfo(code)))
> ...................^
>%CC-I-IMPLICITFUNC, In this statement, the identifier "nl_langinfo" is
> implicitly declared as a function.
> At line number 25 in langinfo.xs.
Sigh, nl_langinfo didn't become available until VMS 6.2, or so the
version dependency tables in the C RTL manual tell me. The following
might do it. Just use your favorite editor to remove the line
preceded by a minus and then add in the lines preceded by pluses
(don't forget to delete the plus signs):
--- configure.com;-0 Tue Jun 11 23:02:33 2002
+++ configure.com Thu Jun 13 14:58:01 2002
@@ -4730,7 +4730,12 @@
$ i_locale="define"
$ i_langinfo="define"
$ d_locconv="define"
-$ d_nl_langinfo="define"
+$ IF vms_ver .GES. "6.2"
+$ THEN
+$ d_nl_langinfo="define"
+$ ELSE
+$ d_nl_langinfo="undef"
+$ ENDIF
$ d_setlocale="define"
$ vms_cc_type="decc"
$ ELSE
[end of possible patch]
Unfortunately that means you'll have to reconfigure and start the
build over from the beginning. You may want to do an MMS REALCLEAN
first.
>The first is an OpenVMS AXP V7.1-1H2 running on a Digital Personal
WorkStation
>(early and slower version of an XP900) with MultiNet V4.1 Rev A, DEC C
V5.7-004,
>and MMS V3.3. Upon the initial run, it had the same error in VMS.C that I
had
>on my VMS 5.5-2 system. Being reasonably intelligent, I tried the same
patch
>on this system and it allowed the make to complete. However, I came up with
8
>test errors herein listed:
>
>t/io/fs..............................FAILED at test 31
>t/op/stat............................FAILED at test 29
>ext/Cwd/t/cwd........................FAILED at test 9
>lib/AutoSplit........................FAILED at test 2
>lib/ExtUtils/t/Command...............FAILED at test 10
>lib/File/stat........................FAILED at test 18
>lib/Net/hostent......................FAILED at test 5
>t/pod/find...........................FAILED at test 2
>
>Failed 8 test scripts out of 628, 98.73% okay.
OK, lib/Net/hostent is a known Multinet problem. There is or soon
will be an ECO from Process. If you have time to get more verbose
output from the others that would be great. The test shell expects
paths relative to the [.t] directory but you see them in your output
relative to the main directory, so you need to run them like:
$ @[.vms]test .exe "" -"v" [.io]fs.t
and
$ @[.vms]test .exe "" -"v" [-.ext.cwd.t]cwd.t
etc.
I don't see any failures from your 7.2-1 system that I don't also see
in your 7.1-1H2 system, so there's no need to look at that
separately. Funny, since I'm getting all green lights on my 7.2-1
and 7.3 systems.
--
________________________________________
Craig A. Berry
mailto:craigberry@mac.com
"... getting out of a sonnet is much more
difficult than getting in."
Brad Leithauser