bundler: 1.14.6 -> 1.16.1 #30881

flokli · 2017-10-27T23:32:13Z

Followup of #30871.

Things done

Tested using sandboxing (nix.useSandbox on NixOS, or option build-use-sandbox in nix.conf on non-NixOS)
Built on platform(s)
- NixOS
- macOS
- other Linux distributions
Tested via one or more NixOS test(s) if existing and applicable for the change (look inside nixos/tests)
Tested compilation of all pkgs that depend on this change using nix-shell -p nox --run "nox-review wip"
Tested execution of all binary files (usually in ./result/bin/)
Fits CONTRIBUTING.md.

manveru · 2017-10-28T12:55:04Z

@zimbatm builds fine for me on NixOS too.

zimbatm · 2017-10-30T13:34:28Z

I'm having issues building nix-build -A mikutter but it's also failing on master. /cc maintainer: @midchildan

zimbatm · 2017-10-30T13:48:07Z

@flokli to you mind changing the base to master? it's not going to be that much to build, ruby packages are quick to build

midchildan · 2017-10-30T17:35:08Z

I'll look into the issue with mikutter. It might take a few days, since I'm kind of busy right now.

flokli · 2017-10-31T16:32:21Z

@zimbatm base is now master again. Feel free to merge :-)

globin · 2017-10-31T16:45:27Z

This will break gitlab probably due to a bug in bundler or our ruby infrastructure, please don't merge until this has been tested!

flokli · 2017-11-15T02:28:47Z

@globin Is this still valid with the latest gitlab version bumps?

flokli · 2017-11-17T10:39:14Z

@globin Updated PR to bundler-1.16.0. I was able to at least build gitlab. Can you give more information on what might possibly break, so I can have a closer look?

flokli · 2017-12-20T00:43:26Z

I rebased this to latest nixpkgs-unstable.

The gitlab service module still seems to be broken on current nixpkgs-unstable - at least the corresponding VM test fails, and I'm unable to get a new gitlab set up via the nixos module either.

So I don't really see how merging this breaks gitlab more than it currently already is.

@globin, @fpletz: Could you please elaborate?

zimbatm · 2017-12-20T10:00:33Z

We should probably remove gitlab from the package set. It's a large piece of software with all sorts of dependencies and ruby omnibus (which is it's own shade of hell). Each dependency change makes the upgrade difficult. Or use the debian package and buildFHS, they already bundle all the dependencies anyways.

Personally I would use the Docker image version since it's shipped by gitlab with timely security updates and has been tested thoroughly.

FRidh · 2017-12-20T10:32:22Z

GitLab is an example of a large project that deserves its own overlay, just like e.g. openstack.

mkaito · 2018-03-03T16:03:04Z

The fact that Gitlab can keep an important (for us Ruby coders) program from being updated in an OS repo is quite frankly unacceptable. Gitlab should have to conform to the Ruby world, not the other way around. If Gitlab can't be made to work, it should be split into an overlay.

Gitlab is a pain to run. I don't know anyone that doesn't just run one of the prepackaged distributions they provide.

Might as well bump to 1.16.1 while we're at it.

manveru · 2018-03-03T16:07:17Z

The alternative is to provide gitlab its own version of bundler?

mkaito · 2018-03-03T17:11:34Z

If the Gitlab derivation can somehow handle it internally without affecting the rest of the repo, that'd work, I suppose.

fpletz · 2018-03-04T08:28:33Z

This is not only a problem with Gitlab but something we've seen with other Rails applications like frab, too. Our Ruby/Bundler infrastructure is somehow broken with this new version. Please fix this first before merging this.

Regarding Gitlab: I agree that we should probably drop Gitlab. It's a big maintenance burden.

zimbatm · 2018-03-05T18:13:17Z

@fpletz @globin do you still use the nixpkgs gitlab?

GrahamcOfBorg · 2018-03-06T15:36:08Z

Success on aarch64-linux (full log)

Partial log (click to expand)

  Gem home: /nix/store/0rz0s2j24244b05mdl22pcsqh4ij0lik-bundler-1.16.1/lib/ruby/gems/2.4.0
Successfully installed bundler-1.16.1
1 gem installed
post-installation fixup
shrinking RPATHs of ELF executables and libraries in /nix/store/0rz0s2j24244b05mdl22pcsqh4ij0lik-bundler-1.16.1
checking for references to /build in /nix/store/0rz0s2j24244b05mdl22pcsqh4ij0lik-bundler-1.16.1...
copying path '/nix/store/6xcfw92dg21hapb2x3c494ihpmnfrsh9-ruby2.4.3-compass-1.0.3' from 'https://cache.nixos.org'...
building '/nix/store/vzndai783na2p9hjabb4sqfpjdcrnxg6-compass-1.0.3.drv'...
created 37 symlinks in user environment
/nix/store/wjhcsgw0zrf7dmkxpv74lzd4pcbhjyix-compass-1.0.3

GrahamcOfBorg · 2018-03-06T15:38:10Z

Success on x86_64-linux (full log)

Partial log (click to expand)

WARNING:  You build with buildroot.
  Build root: /
  Bin dir: /nix/store/fbbvzgri8gmzq1s5q68f6xh87771rhf4-bundler-1.16.1/lib/ruby/gems/2.4.0/bin
  Gem home: /nix/store/fbbvzgri8gmzq1s5q68f6xh87771rhf4-bundler-1.16.1/lib/ruby/gems/2.4.0
Successfully installed bundler-1.16.1
1 gem installed
post-installation fixup
shrinking RPATHs of ELF executables and libraries in /nix/store/fbbvzgri8gmzq1s5q68f6xh87771rhf4-bundler-1.16.1
checking for references to /tmp/nix-build-bundler-1.16.1.drv-0 in /nix/store/fbbvzgri8gmzq1s5q68f6xh87771rhf4-bundler-1.16.1...
/nix/store/fbbvzgri8gmzq1s5q68f6xh87771rhf4-bundler-1.16.1

GrahamcOfBorg · 2018-03-06T15:38:16Z

Success on x86_64-darwin (full log)

Partial log (click to expand)

installing
buildFlags:
WARNING:  You build with buildroot.
  Build root: /
  Bin dir: /nix/store/fclqz0j2qhdn9nbw5s8crr1l8lpz2waq-bundler-1.16.1/lib/ruby/gems/2.4.0/bin
  Gem home: /nix/store/fclqz0j2qhdn9nbw5s8crr1l8lpz2waq-bundler-1.16.1/lib/ruby/gems/2.4.0
Successfully installed bundler-1.16.1
1 gem installed
post-installation fixup
/nix/store/fclqz0j2qhdn9nbw5s8crr1l8lpz2waq-bundler-1.16.1

GrahamcOfBorg · 2018-03-06T15:38:20Z

Success on aarch64-linux (full log)

Partial log (click to expand)

/nix/store/0rz0s2j24244b05mdl22pcsqh4ij0lik-bundler-1.16.1

Mic92 · 2018-03-12T18:52:26Z

Having a different bundler in ruby.devEnv and nixpkgs breaks applications that uses bundler such as vagrant. This is unacceptable and should be addressed.

vcunat · 2018-03-12T19:07:28Z

This isn't bugfix-only, right? (and it seems risky) Why do you think it's worth rushing it into 18.03?

Mic92 · 2018-03-12T19:11:06Z

@vcunat because currently other applications are broken because of this.

Mic92 · 2018-03-12T19:14:54Z

I am not sure, if the bundler in ruby.devEnv could be downgraded though. This would also fix the problem.

Mic92 · 2018-03-12T19:24:52Z

According to the bundler homepage using rubygems 2.2 requires bundler 1.6.1: https://bundler.io/compatibility#bundler-compatibility-with-rubygems (for c-extensions)
Update maybe not in our setup.

Mic92 · 2018-03-12T21:08:25Z

Workaround for vagrant: #36880

polynomial · 2018-03-22T07:14:17Z

I have a team that currently can't use 18.03 because of this. What are the next steps to getting this addressed?

mkaito · 2018-03-22T09:06:59Z

We use overlays for this kind d of thing. El 22 de marzo de 2018 7:14:22 GMT+00:00, Benjamin Smith <notifications@github.com> escribió:

…

I have a team that currently can't use 18.03 because of this. What are the next steps to getting this addressed? -- You are receiving this because you commented. Reply to this email directly or view it on GitHub: #30881 (comment)

-- Enviado desde mi dispositivo Android con K-9 Mail. Por favor, disculpa mi brevedad.

Mic92 · 2018-03-22T10:04:49Z

@polynomial see what breaks in our bundler infrastructure, I guess.

mguentner · 2018-04-19T20:14:37Z

Any news on this? 💎 😭

My development workflow broke because of that. I am stuck at 17.xx for that reason.

mkaito · 2018-04-20T15:01:22Z

I put this in a private overlay, and it was causing recompiles of all kinds of things, down to libgtk. I have no idea how bundler could possibly cause libgtk to recompile, but removing the overlay fixed it.

I think the policy here should be to prefer updating core modules like bundler, and let dependants figure it out. Especially big bad stuff like Gitlab shouldn't cause this to be held back.

I run into constant "you've already activated bundler 1.16.x but are trying to activate 1.14.x" while trying to work, even without that overlay.

manveru · 2018-04-20T15:57:16Z

Generally, please use the bundlerEnv{}.wrappedRuby and also bundlerEnv{}.bundler instead of the global nixpkgs ones. That should alleviate 99% of the problems.
Edit: also, never run bundler exec, it's evil!

mkaito · 2018-04-20T18:44:33Z

I usually load ruby_2_5.devEnv via nix-shell. That would put the correct things on the path? El 20 de abril de 2018 16:57:19 GMT+01:00, Michael Fellinger <notifications@github.com> escribió:

…

Generally, please use the `bundlerEnv{}.wrappedRuby` and also `bundlerEnv{}.bundler` instead of the global nixpkgs ones. That should alleviate 99% of the problems. -- You are receiving this because you commented. Reply to this email directly or view it on GitHub: #30881 (comment)

-- Enviado desde mi dispositivo Android con K-9 Mail. Por favor, disculpa mi brevedad.

mguentner · 2018-04-21T16:13:09Z

@manveru

Thanks for the hint with wrappedRuby. However this still gives me

/nix/store/l3lbs5jfxg215hxnvi5rlf3ymyxzf2xf-bundler-1.14.6/lib/ruby/gems/2.4.0/gems/bundler-1.14.6/lib/bundler/runtime.rb:40:in `block in setup': You have already activated bundler 1.16.1, but your Gemfile requires bundler 1.14.6. Prepending `bundle exec` to your command may solve this. (Gem::LoadError)

Do you have a working development derivation that works w/ the current master/unstable and bundix?

Also +1 to @mkaito - but priorities tend to shift if you happen to use "big bad stuff" yourself.

@globin 👀

manveru · 2018-04-21T17:41:22Z

For that I have no simple solution siche since it seems that your Gemfile itself had a hard version lock... Any chance you could remove that line from it?

…

On Sat, 21 Apr 2018, 18:13 Maximilian Güntner, ***@***.***> wrote: @manveru <https://github.com/manveru> Thanks for the hint with wrappedRuby. However this still gives me /nix/store/l3lbs5jfxg215hxnvi5rlf3ymyxzf2xf-bundler-1.14.6/lib/ruby/gems/2.4.0/gems/bundler-1.14.6/lib/bundler/runtime.rb:40:in `block in setup': You have already activated bundler 1.16.1, but your Gemfile requires bundler 1.14.6. Prepending `bundle exec` to your command may solve this. (Gem::LoadError) Do you have a working development derivation that works w/ the current master/unstable and bundix? Also +1 to @mkaito <https://github.com/mkaito> - but priorities tend to shift if you happen to use "big bad stuff" yourself. @globin <https://github.com/globin> 👀 — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#30881 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAANs-qqmjRPUposfKkyPVUkWktF4-mmks5tq1qYgaJpZM4QJuia> .

mguentner · 2018-04-21T17:54:05Z

Frankly, I have no idea what is going on. Even if I remove the version lock, the error stays the same.

Mic92 · 2018-04-22T23:30:18Z

@globin could you just push your local fix to some branch, we worked on for gitlab, so that we can continue with this update?

globin · 2018-04-26T15:44:44Z

The gitlab changes have been pushed and should work with this PR now

Mic92 · 2018-04-26T17:04:48Z

I will take a look at #37611 instead.

flokli force-pushed the bump-bundler branch from 140e9ef to 64c137b Compare October 31, 2017 16:26

flokli requested a review from FRidh as a code owner October 31, 2017 16:26

flokli changed the base branch from staging to master October 31, 2017 16:27

flokli force-pushed the bump-bundler branch from 64c137b to ca41bac Compare October 31, 2017 16:29

GrahamcOfBorg added 10.rebuild-darwin: 11-100 10.rebuild-linux: 11-100 labels Oct 31, 2017

midchildan mentioned this pull request Nov 1, 2017

mikutter: 3.5.7 -> 3.5.13 #31074

Merged

8 tasks

flokli force-pushed the bump-bundler branch from ca41bac to 373c088 Compare November 17, 2017 10:37

flokli requested a review from zimbatm as a code owner November 17, 2017 10:37

flokli changed the title ~~bundler: 1.14.6 -> 1.15.4~~ bundler: 1.14.6 -> 1.16.0 Nov 17, 2017

GrahamcOfBorg added the 10.rebuild-linux: 101-500 label Nov 17, 2017

flokli force-pushed the bump-bundler branch from 373c088 to 1577022 Compare December 20, 2017 00:30

FRidh removed their request for review December 20, 2017 10:31

FRidh mentioned this pull request Feb 27, 2018

bundler: 1.14.6 -> 1.16.1 #35834

Closed

8 tasks

GrahamcOfBorg added 10.rebuild-linux: 501+ and removed 10.rebuild-linux: 101-500 10.rebuild-linux: 11-100 labels Mar 6, 2018

Mic92 added this to the 18.03 milestone Mar 12, 2018

Mic92 mentioned this pull request Mar 12, 2018

bundler: 1.14.6 -> 1.16.1 #36409

Merged

8 tasks

xeji mentioned this pull request Mar 22, 2018

bundler 1.14.6 -> 1.16.1 #37611

Closed

8 tasks

Mic92 closed this Apr 26, 2018

bundler: 1.14.6 -> 1.16.1 #30881

bundler: 1.14.6 -> 1.16.1 #30881

Conversation

flokli commented Oct 27, 2017 • edited

Things done

manveru commented Oct 28, 2017

zimbatm commented Oct 30, 2017

zimbatm commented Oct 30, 2017

midchildan commented Oct 30, 2017

flokli commented Oct 31, 2017

globin commented Oct 31, 2017

flokli commented Nov 15, 2017

flokli commented Nov 17, 2017 • edited

flokli commented Dec 20, 2017

zimbatm commented Dec 20, 2017

FRidh commented Dec 20, 2017

mkaito commented Mar 3, 2018 • edited

manveru commented Mar 3, 2018

mkaito commented Mar 3, 2018

fpletz commented Mar 4, 2018

zimbatm commented Mar 5, 2018

GrahamcOfBorg commented Mar 6, 2018

GrahamcOfBorg commented Mar 6, 2018

GrahamcOfBorg commented Mar 6, 2018

GrahamcOfBorg commented Mar 6, 2018

Mic92 commented Mar 12, 2018

vcunat commented Mar 12, 2018

Mic92 commented Mar 12, 2018 • edited

Mic92 commented Mar 12, 2018

Mic92 commented Mar 12, 2018 • edited

Mic92 commented Mar 12, 2018 • edited

polynomial commented Mar 22, 2018

mkaito commented Mar 22, 2018 via email

Mic92 commented Mar 22, 2018

mguentner commented Apr 19, 2018 • edited

mkaito commented Apr 20, 2018

manveru commented Apr 20, 2018 • edited

mkaito commented Apr 20, 2018 via email

mguentner commented Apr 21, 2018

manveru commented Apr 21, 2018 via email

mguentner commented Apr 21, 2018

Mic92 commented Apr 22, 2018

globin commented Apr 26, 2018

Mic92 commented Apr 26, 2018

flokli commented Oct 27, 2017 •

edited

flokli commented Nov 17, 2017 •

edited

mkaito commented Mar 3, 2018 •

edited

Mic92 commented Mar 12, 2018 •

edited

Mic92 commented Mar 12, 2018 •

edited

Mic92 commented Mar 12, 2018 •

edited

mguentner commented Apr 19, 2018 •

edited

manveru commented Apr 20, 2018 •

edited