Comment: DomainKeys? See http://antispam.yahoo.com/domainkeys
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws;
  s=default; d=rowman.com;
  b=XjlHzBzvgUW5rczLrCpebKJmqg/v0Ww75KbpDLDFBzDWPe562yzeJ7Q/l3plqfjc6DrOiha9kzvk80m5kJSwBiGLXOe5HZ18czDUIYI+f4XYDTwxyA9Vjw2XSzIE4yd/pKpXz9zlxDRE9hU2VQX4hprPkG3f4dgQfEr7IPQkLSA=  ;
Message-ID: <45577075.2090408@rowman.com>
Date: Sun, 12 Nov 2006 14:05:25 -0500
From: John Peacock <jpeacock@rowman.com>
User-Agent: Thunderbird 1.5.0.7 (X11/20060911)
MIME-Version: 1.0
CC: "Andreas J. Koenig" <andreas.koenig.gmwojprw@franz.ak.mind.de>,
        module-build@perl.org
Subject: Re: MB 0.2805/version.pm broken on older bleadperls
References: <87d57sligo.fsf@k75.linux.bogus> <45575A46.6030908@rowman.com>
In-Reply-To: <45575A46.6030908@rowman.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Lines: 44
Xref: k75.linux.bogus king-2002:46766

John Peacock wrote:
> I backrev'd my p4 checkout to that rev and I'll try to build bleadperl and then
> get Module::Build to play nice if I can.  It's kind of a stretch, though, to try
> and maintain compatibility with random bleadperl releases, which by default and
> design are not intended to be considered stable releases.

OK, here's the story:  back in 2005-06-06, the version object was much simpler
(it was just a blessed array).  And $^V returned one of those simpler version
objects at that time, which Module::Build's version.pm code has no clue how to
deal with (it expects version objects to be hashes, not arrays).  We can special
case this in the Module::Build code, but I'm really not sure it's worth putting
it in the released code, since bleadperl has moved on from that time in an
incompatible fashion.

For your testing purposes only, apply the following patch:

=== lib/Module/Build/Base.pm
==================================================================
--- lib/Module/Build/Base.pm    (revision 1980)
+++ lib/Module/Build/Base.pm    (local)
@@ -1168,7 +1168,7 @@
   # Check the current perl interpreter
   # It's much more convenient to use $] here than $^V, but 'man
   # perlvar' says I'm not supposed to.  Bloody tyrant.
-  return $^V ? $self->perl_version_to_float(sprintf "%vd", $^V) : $];
+  return $];
 }


which will avoid the test that is freaking out.  This could be applied to the
M::B core (as $] and $^V will produce equivalent version objects), but I
wouldn't recommend it just to support a now-expired bleadperl...

John

-- 
John Peacock
Director of Information Research and Technology
Rowman & Littlefield Publishing Group
4501 Forbes Blvd
Suite H
Lanham, MD 20706
301-459-3366 x.5010
fax 301-429-5747