Conversation

Notices

1. Embed this notice
   Miah Johnson (miah@hachyderm.io)'s status on Monday, 23-Jan-2023 09:47:29 JST Miah Johnson

   Had a thought earlier that the reason containers gained so much traction quickly is deploying apps written in dynamic languages tends to be a terrible experience with millions of files and dependencies.

   You don't really have the same issue deploying any app that has a compiled output.

   Dropping a .jar in a directory and running it is trivial compared to deploying deploying a [Ruby,Python,Node] app.

   • Embed this notice
     Kit Rhett Aultman (roadriverrail@signs.codes)'s status on Monday, 23-Jan-2023 09:47:28 JST Kit Rhett Aultman

     in reply to

     @miah Yes. IMHO, it also tracks with the explosion in dev environments that intentionally outpace Linux distros. Forcing devs to break from the regularity of the distro ultimately made "IDK, just make a Linux entirely for the app" necessary.

     That said, as a firmware dev, containers are great for keeping regularity of build environment across my team without forcing us to all use a single outdated Linux distro because some vendor wants that. So, yeah, I'm a hypocrite.

   • Embed this notice
     Kit Rhett Aultman (roadriverrail@signs.codes)'s status on Tuesday, 24-Jan-2023 05:45:36 JST Kit Rhett Aultman

     in reply to

     • グレェ「grey」

     @byterhymer @miah These are all incredibly cogent and valuable insights. As a firmware dev, I generally don't have to think about ops/production in that way. Instead, I mostly keep scratching my head, because I grew up and was educated in a world where every computing resource was precious, and now I'm in a world where "just put a machine inside the machine inside the machine" is a legit answer for most everything. And it's confusing, but only in an abstract way.

   • Embed this notice
     グレェ「grey」 (byterhymer@mastodon.social)'s status on Tuesday, 24-Jan-2023 05:45:37 JST グレェ「grey」

     in reply to

     • Kit Rhett Aultman

     @roadriverrail IMHO, that tracks. Personally I find that VMs/etc. are EXCEEDINGLY USEFUL for dev.

     By the same token: they are the OPPOSITE of what I ever want in prod.

     For CI frameworks & such? Yes, please: I will take performance hits for some ease of reproducibility on recreating bare build environments to make sure codes' assumptions are doing what they need to be doing.

     Alas, too many took the "ease of dev" into Ops & there were & are GOOD REASONS to not EVER cross those streams.
     @miah