Support more minitar versions & prep for r10k 5.0 #major #1408
Conversation
This changes us from requiring at least minitar 1.0 and instead goes to the previous minimum verion (0.9), but instead of using a "~>" which forbids going to a new major release, ie is equivalent to [">= 0.9", "< 1.0"] we simply require above the previous version, not caring if we take up a new major release. This allows us to be installed in the same version Ruby bundle as Puppet (which currently does not support minitar >= 1.0). Minitar has supported the "new" `::Minitar` constant since before 0.9, so there are no code changes required.
|
Do not merge yet. I definitely want @bastelfreak 's feedback first! |
| - Add Ruby 3.3 to CI [#1403](https://github.com/puppetlabs/r10k/pull/1403) | ||
| - Require Ruby 3.1 [#1402](https://github.com/puppetlabs/r10k/pull/1402) | ||
| - Support minitar gem > 1.0. [#1407](https://github.com/puppetlabs/r10k/pull/1407) |
There was a problem hiding this comment.
shoot, this PR ended up being 1408, I guess I missed 1407.
There was a problem hiding this comment.
To prevent this in the future I raised #1404 some time ago :D
Test for puppetlabs/r10k#1408
|
I gave this a quick test on ruby 3.1 and newer + puppet_forge 6 / minitar 1.0.2 in our downstream tool ra10ke: voxpupuli/ra10ke#110 . That looks fine. Edit: |
Co-authored-by: Tim Meusel <[email protected]>
Co-authored-by: Tim Meusel <[email protected]>
|
How do you feel about making this a 5.0 release, @bastelfreak ? Are there any other PRs you'd like to see merged for a 5.0 release? |
|
Can you take a look at #1407 about the gettext integration? In case that's dead we could rip it out now as well. |
|
I looked into removing gettext and unfortunately we can't at the moment. I'm still open to other PRs getting in for 5. I am unfortunately taking Friday and next week off. It feels rude to push out this release another two weeks, but it feels more irresponsible to push out a release and then leave for a week. While I'm gone please let me know what you think of existing PRs and open any other small changes you'd like to go out! |
|
I totally agree that we should not rush major releases. I don't have anything else in my mind for the next release right now, but I will think about it. |
|
Hey Tim, Seems like there was changes while I was on break! I've gotten approval to go ahead and keep releasing R10K though. Have you thought about additional changes to go in? |
|
hi, |
|
@justinstoller @bastelfreak what about #1371? It would be good to fix the 'bad object ref' issue and the change could be considered breaking. |
|
I created #1410 following @bastelfreak's suggestion to make a separate PR for the pruning issue. Let me know if there's anything else I can do. |
|
See #1411 for the non-release part of this PR. I'm going to merge that while we figure out if there's a simple fix to #1399 (comment) or if it's something we shouldn't block the release on. |
|
@justinstoller I won't have time to look at #1399 until after the US holiday break just so you're aware of the timing. |
|
The last of these changes, the 5.0.0 release part, is now in #1413 |
I would consider moving from requiring Ruby 2.6 to Ruby 3.1 to be a major breaking change. One we should totally do, but one I think requires a major version bump. We should see what else in the PR backlog we want to get in for such a release.
Also, this allows us to support minitar version > 0.9, to smooth transitions for people who require both r10k and puppet in their Gemfiles/gemspecs.