On Tue, Aug 26, 2014 at 11:57 PM, Daniele Sluijters <
[email protected]> wrote:

> Hey,
>
> I agree with the spirit of the fix but the fact that it isn't mentioned
> anywhere in the release notes is a bit annoying.
>

Yep, it's a release notes fail, and it happened because we didn't track the
change with separate tickets. What happened is we made this change as part
of adding lsbminordistrelease (FACT-637
<https://tickets.puppetlabs.com/browse/FACT-637>) and adding the new 'os'
structured fact (FACT-614) <https://tickets.puppetlabs.com/browse/FACT-614>,
but it wasn't called out separately. Anyway that's an explanation, but not
really an excuse - we dropped the ball on publicizing this change.

We've added tickets for the changes and pushed updated release notes for
facter 2.2 here <https://docs.puppetlabs.com/facter/2.2/release_notes.html>.


> I personally also consider this a backwards incompatible release, you're
> changing old behaviour. Albeit for the better, but people depended on that
> behaviour and no prior warning or deprecation warning was issued.
>

Although we clearly should have documented this differently, since it's in
the wild and there's consensus that this change *is* for the better, we'd
like to leave it be.

But going forward there's a question about how to handle changes to fact
*values*. One proposal is that we identify (and of course test against)
some essential facts that we "care a lot about" (such as
'lsbmajdistrelease") and set some rules, like:

(a) we do not change those in x.y.Z releases
(b) we highlight it when they DO change in x.Y or X releases


> Do also keep in mind that though we can easily fix our own manifests with
> a regexp match, since the launch of the Forge people have started to use
> more and more modules maintained by others. Unless the maintainer has
> already issued an update you're stuck with manually patching an upstream
> module and carrying that change. Depending on how you deploy your
> environment this might be difficult to do.
>

I grep'd my way through forge modules looking for affected modules and
there are actually just a handful (3 related to postgresql, and 3 others).
I'll ping those authors to let them know. There's already a fix in for
puppetlabs-postgresql.

Btw, one last thing: another plug for the recently announced
<https://groups.google.com/forum/#!topic/puppet-users/gtXrUgQZmJ0> nightly
repos, which would have exposed this issue if we'd had repos all in place
ahead of time (this change went in two weeks before release).

Thanks!

-- 
Kylo Ginsberg
[email protected]

*Join us at PuppetConf 2014 <http://www.puppetconf.com/>, September
20-24 in San Francisco*
*Register by September 8th to take advantage of the Final Countdown
<https://www.eventbrite.com/e/puppetconf-2014-tickets-7666774529?discount=FinalCountdown>
*
*—**save $149!*

-- 
You received this message because you are subscribed to the Google Groups 
"Puppet Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/puppet-users/CALsUZFHQ_Pho_%2Bq9VCBMdMhe4DjcZRhvmF5NU74wTU-DXoZ9xg%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to