weixin_39969953
weixin_39969953
2020-12-09 12:37

Sleep to deprecation notices

Inspired from a blog post about using sleep in compiling C++ programs this RFC will add sleep to our deprecation notices. This is a way for us to gently nudge our users to pay attention to the notices so they aren't effected in major releases.

Signed-off-by: JJ Asghar

该提问来源于开源项目:chef-boneyard/chef-rfc

  • 点赞
  • 写回答
  • 关注问题
  • 收藏
  • 复制链接分享
  • 邀请回答

10条回答

  • weixin_39673051 weixin_39673051 5月前

    👎 on this one too

    点赞 评论 复制链接分享
  • weixin_39954674 weixin_39954674 5月前

    👎 as well

    点赞 评论 复制链接分享
  • weixin_39717367 weixin_39717367 5月前

    different problem space though and on production servers where you want chef-client runs to be finishing in 60 seconds or less this can impact your ability to deploy changes.

    i'd also find that behavior of a compiler completely infuriating and the first thing i'd do would be to add whatever environment variable to my .zshrc that would make that go away.

    点赞 评论 复制链接分享
  • weixin_39522927 weixin_39522927 5月前

    Yeah, I’m -1 on this as well, even limited to just Test Kitchen. Because of the complex web of community cookbooks, many warnings are not actionable by the person seeing them. This is the same reason that warnings as errors isn’t on by default in Kitchen.

    点赞 评论 复制链接分享
  • weixin_39762666 weixin_39762666 5月前

    granular warnings as errors would be cool, if you could extend policyfile/berks to let you know which cookbooks are likely owned by the user.

    点赞 评论 复制链接分享
  • weixin_39522927 weixin_39522927 5月前

    Part of the structlog context work on my list is the ability to do per cookbook debug logging, hopefully we could do a similar thing for warnings too.

    点赞 评论 复制链接分享
  • weixin_39518678 weixin_39518678 5月前

    I think this is a nice approach for things like the ChefDK and InSpec CLI where there's interaction with a human, but I'm :-1: on this being used with CCRs.

    点赞 评论 复制链接分享
  • weixin_39969953 weixin_39969953 5月前

    Turns out it wasn't an article I found on hacker news but this tweet:

    https://twitter.com/johnregehr/status/920691341738123264

    It get's the point across too though :)

    点赞 评论 复制链接分享
  • weixin_39717367 weixin_39717367 5月前

    After seeing the tweet, I'm a very solid 'no' 👎 on this in general. Either with CCRs or with the CLI. In most cases it is just going to be an abusive UX experience to the user.

    There's a very specific case where the problem is something like that deprecated library where all the user needs to do is delete linking against it where I can see it.

    But this was not being added to -Wall notices from the compiler or anything like that.

    点赞 评论 复制链接分享
  • weixin_39624700 weixin_39624700 5月前

    Yeah, I think this not something we want to do as part of a CCR. Building better tooling in Automate and so on to surface deprecations is much less user intrusive.

    Thanks for the RFC,

    点赞 评论 复制链接分享

相关推荐