NixOS package missing version and git commit

[abiosoft]

Describe the Issue

Running colima version on nixpkg shows wrong version and git commit.

Version

N/A

Colima Version:

What is the output of colima version

colima version development
git commit: unknown

Lima Version:

What is the output of limactl --version

N/A

Qemu Version

What is the output of qemu-img --version

N/A

Operating System

• macOS Intel
• macOS m1
• Linux

To Reproduce

Steps to reproduce the behavior:

1. Install Colima via nixpkgs
2. Run colima version

Expected behavior

Colima shows version and revision

Additional context

cc @aaschmid @Atemu

[abiosoft]

@aaschmid re #301 (comment), are there any file permission restriction on the environment? That is among the likely causes.

The test is basically testing a background daemon and creates a pid file in the process. I was trying to mimic an actual use as much as possible.

Anyways, I am already looking at stubbing it out as it seems to be an issue here.

[aaschmid]

I am not aware of any special permissions, though.

In addition, the PR for v0.4.2 (see NixOS/nixpkgs#174013) does also have failed build, see e.g. https://github.com/NixOS/nixpkgs/pull/174013/checks?check_run_id=6545228494.

[aaschmid]

As a workaround we could just disable the tests for now... (ugly but effective to get the new version available).

[abiosoft]

That is fine by me.

But how easy would it be to re-enable it? I would definitely be looking to get it fixed before the next release.

[abiosoft]

I made a new change that should simply skip the test on failure. You can try the branch once again and see.
If it passes this time, I can push a new release just for this. If otherwise, let us skip the tests for this version.

Thanks.

[Atemu]

This issue seems to be about the upgrade to the newest version. The missing version is already fixed by NixOS/nixpkgs#173914.

It's better to track these kinds of issues in Nixpkgs though. This isn't related to this app's code, just our packaging of it.

[abiosoft]

I finally reproduced this error #301 (comment). It required me setting up Nix and running in a pure nix shell nix-shell --pure with this shell.nix.

Not a big issue and tests can be skipped for v0.4.2. It will be properly addressed ahead of future releases.

PS: current master should no longer fail the tests.

[aaschmid]

PS: current master should no longer fail the tests.

You mean main from 4 days ago?

[abiosoft]

You mean main from 4 days ago?

Oops yeah, I meant main.

[aaschmid]

Test seems to be skipped but now on my machine the following fails:

=== RUN   Test_config_Mounts
=== RUN   Test_config_Mounts/0
time="2022-05-29T18:56:33Z" level=fatal msg="cannot make required directory: mkdir /homeless-shelter: read-only file system"
FAIL    github.com/abiosoft/colima/environment/vm/lima  0.264s
FAIL

Where is the directory try to be created?

P.S.: the output for the skipped test

ok      github.com/abiosoft/colima/cmd  0.327s
=== RUN   TestStart
    daemon_test.go:31: pidfile /private/tmp/nix-build-colima-0.4.2.drv-0/TestStart2548514758/001/daemon.pid
    daemon_test.go:46: start successful
    daemon_test.go:53: daemon not supported: context deadline exceeded
--- SKIP: TestStart (5.01s)
PASS

[abiosoft]

This has been fixed for 0.3.4. after upgrading to Nix 2.8.1.

$ nix --version
nix (Nix) 2.8.1

$ nix shell nixpkgs#colima

$ colima version
colima version 0.3.4
git commit: 5a4a70481ca8d1e794677f22524e3c1b79a9b4ae

$ which colima
/nix/store/nygq9gxi21b939dyvkdg2w0f5xix7gng-colima-0.3.4/bin/colima

@aaschmid should we close the issue and track the failing tests on nixpkgs repo?

[aaschmid]

We can track the failing tests on nixpkgs repo also, yes. As you prefer, I am flexible ;-)
Thanks for your effort anyways!

[abiosoft]

@aaschmid as a side node, what resources do I use to keep up with Nix?

I gave Nix a second try due to this issue (tried it before, gave up and reverted to Brew) and I think I finally got a hang of it. Only to check out flakes which led me to 2.0 commands, and now I am totally lost.

My use case was generally covered with nix-shell, nix-shell -p ... and home-manager. But now there's a million and one new commands that sounds like they all do the same thing. 😕

I keep seeing flakes everywhere and sounds like it is the direction Nix is heading, but there is not as much documentation compared to what I am used to with the tools I generally use. And much of docs assume you're already a Nix expert.

Anyways, enough of the rant. Closing the issue.

[Atemu]

https://discourse.nixos.org/c/announcements/8 is a pretty good place.

The new stuff with the new nix CLI, flakes etc. is still experimental and subject to change. If you use-cases are covered by nix-shell and hm, that's absolutely fine and you should keep using them. That's how my usage looks like too.

Official and non-official docs are sparse but if you have any suggestions on the official ones, feel free to open issues on Nix and Nixpkgs.

[aaschmid]

The new stuff with the new nix CLI, flakes etc. is still experimental and subject to change. If you use-cases are covered by nix-shell and hm, that's absolutely fine and you should keep using them. That's how my usage looks like too.

I totally agree with @Atemu.
And to be honest, I steered clear of flakes for a while now. This is becausae of a available time reason but first of all the not found obvious benefit / requirement of flakes for me.

keep seeing flakes everywhere and sounds like it is the direction Nix is heading, but there is not as much documentation compared to what I am used to with the tools I generally use. And much of docs assume you're already a Nix expert.

Yes, I am quite familiar with the usage of nix, nix-darwin and home-manager but the learning curve was rather steep. For the creation of packages there is a bunch of additional things to learn which I try on demand - and I always learn new things ;-)