PPM Builds

Hello list,

Hopefully this is the right place to discuss this - not the
PPM tool
itself, but the builds done by ActiveState to populate
their
repository.  If this is the wrong place, and there is a
right place,
please let me know.

I track the ActiveState builds of Imager for QC, though
sometimes the
information returned has been less than useful.

At the moment the build results for Imager 0.55 on 5.8 are
puzzling:

- the software being built on 5.8 seems to be a mix of 0.53
and 0.55,
for example the i_rubthru() function was moved from image.c
to
rubthru.im (pre-processed to rubthru.c) in 0.54, but the
build
complains it's defined in both.  Some of the warnings listed
are for
comment lines.

Is there any way to track what has caused this?

Imager seems to be building ok elsewhere:

http://cpantesters.perl.org/show/Imager.html#Imager-0.55


-- 
Tony
_______________________________________________
PPM mailing list
PPMlistserv.ActiveState.com

http://listserv.ActiveState.com/mailman/listinfo/ppm

On Wed, Jan 17, 2007 at 10:06:04AM -0500, D D Allen wrote:
> While I'm not sure that this has anything to do with
the problem, you 
> should be aware that ActiveState intentionally uses
very old versions of 
> Perl in their PPM build process.  For example,
ActiveState uses Perl 5.8.0 
> to build 5.8.x PPMs for Windows.   And ActivePerl 5.8.0
dates to March 
> 2003... coming up on 4 years old.   You can see this on
the ActiveState 
> "Build Status" page for Imager --> 
> http://ppm.activestate.com/BuildSt
atus/5.8-windows/windows-5.8/Imager-0.55.txt

The older version of perl doesn't bother me - I test on perl
back to
5.005_03.

> I've always found Philippe M. Chiasson <gozerActiveState.com> to be 
> highly responsive especially to CPAN authors.  If he
hasn't responded to 
> your email already, I would suggest that you send him
an email asking for 
> help diagnosing the PPM build problems.

Thanks, I'll follow up with Philippe.

> An alternative would be to ask Randy Kobes to build and
host an Imager PPM 
> (for Windows) on the University of Winnipeg's
"Theory58s" repository. 
> Randy's support for the Perl community is outstanding
and my experience is 
> that the UWinnipeg repository can host many ActiveState
PPMs for CPAN 
> distributions that fail ActiveState's build process. 
You generally need 
> only to send an email to Randy asking him to build and
host a Windows PPM 
> for your CPAN distribution.  See the following --> 
> http://theoryx5.uw
innipeg.ca/ppms/

I host a Win32 PPM for 5.8 myself, and there's a few other
places that
host varying builds.

Thanks for the help.

Tony
_______________________________________________
PPM mailing list
PPMlistserv.ActiveState.com

http://listserv.ActiveState.com/mailman/listinfo/ppm

On Fri, Jan 19, 2007 at 01:43:08PM -0800, Philippe M.
Chiasson wrote:
> Sorry for the slight delay in responding to this, but I
wanted to wait for
> actual good news before responding.
> 
> The short story is that, indeed, the PPM build system
was working off a messed
> up version of Imager, with bits from an older version
mixed up with a newer version.
> This was caused by some strange bug in the CPAN
importing component.
> 
> It's been fixed, and I pushed thru a rebuild of Imager
on all platforms.
> 
> htt
p://ppm.activestate.com/BuildStatus/5.8-I.html

Thanks for that, now I just need to figure out why the HP
builds are
failing.

One of them looks strange:

http://ppm.activestate.com/Bui
ldStatus/5.8-hpux/hpux-ia64-lp64-5.8/Imager-0.55.txt

complains about imicon.o, but nothing in the log builds it
(and it
should have.)

The other problems seem to be Imager's Makefile.PL replacing
C<
-L/usr/lib/hpux64 > with C< -L/usr/lib/hpux32 >,
though I can't see
where it's coming from, unless it's in libpth or loclibpth.

I don't see a build log for hpux-pa-risc.

I guess it's time to fire up my old TestDrive account.

Tony

_______________________________________________
PPM mailing list
PPMlistserv.ActiveState.com

http://listserv.ActiveState.com/mailman/listinfo/ppm