Prepare the 1.0.0 release.

[apenney]

No description provided.

[hunner]

So, it looks like there were feature adds during the 1.0.0-rc1 release (adding Archlinux) which is technically not supposed to be done. Or was it a bugfix? We will have to back that out before doing a 1.0.0 release to stick to semver, or call it a learning experience and do a 1.0.0-rc2

[apenney]

oh no, arch was always supported, it just gained $osfamily support so we shifted the params up into the main body. It was just a change to match on osfamily rather than operatingsystem

[hunner]

Okay. Bugfixes are fine. Want to make this 1.0.0-rc2 instead then?

[apenney]

I'm torn here. I mean it doesn't change any behavior other than changing where in the case statement it matches. It's the same parameters and same resources, just a different location in a case. Does that count as a bugfix? It worked fine before.

I feel like it's almost a noop change. I get what you're saying but a second rc isn't going to be any more tested than -rc1 one was. If I make this -rc2 I'm still going to want to release a 1.0 really soon after.

[hunner]

Looks like semver doesn't explicitly state the progression rules for pre-releases, so an rc2 isn't strictly necessary I guess.

[apenney]

\o/

[nanliu]

Recommend documentation of Archlinux requiring Facter 1.7.0. I believe you should bump semver, because this can break for older clients.

[apenney]

I guess that is technically true. I was thinking that we're going from 0.0.3 to 1.0.0 and that nobody using arch would have upgraded to 1.0.0-rc1 but I suppose it's potentially possible! The whole Semver thing confuses me.

Does this deserve a 1.1? Do rc's factor in? Because in my mind this is going from 0.0.3 to 1.0.0 which is a semver bump. I guess the lesson here is never merge in anything but tiny changes in an rc. I would assume rc's are outside however and are just candidates, not actual versions and so this really just went from 0.0.3 to 1.

[hunner]

Thanks for catching this @nanliu ! Looks like it wasn't just a bugfix and doing a 1.0.0-rc.2 would have been prudent after all.

We'll back the archlinux change out and do 1.1.0

[hunner]

Looking again, depending on facter 1.7.x from 1.6.x is covered by Semver as a backwards-compatible change.

Additionally, I'm going to invoke this sentence from Semver 2.0.0 section 9 and claim that the changes introduced are covered up until the commit that is tagged 1.0.0: "a pre-release version indicates that the version is unstable and might not satisfy the intended compatibility requirements as denoted by its associated normal version."

Will try to only introduce bugfix changes in pre-release versions in the future and will take this as a learning experience. Thanks for your help!

[hunner]

Also, if modules could express facter/puppet version dependencies, this would have not been an issue :). (Yes, we want that too.)