Andreas J. Koenig wrote:
On Wed, 16 Dec 2009 09:14:40 +0000, "Chris 'BinGOs' Williams"
<ch...@bingosnet.co.uk> said:
> Regarding the email subject, as far as I can tell from the reports I looked
through they
> were all relating to time issues with Makefile, CP::D::Build wouldn't be involved
> CP::D::MM would.
Sorry for the bad subject line and thank you for the fine grained
details.
Do you have suggestions how we can prevent this from happening again?
Are the timestamps of the untarred distributions in the future on such
systems? Presumably this would play an ugly game also on CPAN.pm based
smokers, right?
Yeah, I rsync from the upstream mirror ( cpan.dagolden.com ) every
hour and fire off the smokers for new dists. My FreeBSD box after a few
days would lag terribly, losing several minutes per hour! The skew got
so bad, what you explained happened - untarred distros were in the
future. I assume it would affect any toolchain modules trying to
install/smoke them?
I wonder how it would affect a "normal" CPAN user trying to install
the modules on a system? Is it possible for EU::MM/CP:D:MM/friends to
automatically detect this "funkiness" and re-run the makefile, so
nothing will blow up if this situation happens on a server? I'm willing
to test this theory and investigate possible ways to patch the toolchain
to do this, but only if somebody tells me it's feasible :)
I've been busy having christmas potlucks with friends/family, and
once I have some free time I'll get NTPd running on the boxes and watch
them closely for any clock skew issues. Maybe I can add a filter to my
outbound mailer to discard reports if it sees the "makefile out of date"
message and send alerts to me... :)
Thanks again!
