`ruby-ansi` failure on 'staging'

`ruby-ansi` failure on 'staging'

[Marius Bakke]

[-- Attachment #1: Type: text/plain, Size: 962 bytes --]

On the 'staging' branch (with ruby 2.4), the 'ruby-ansi' package fails
during the 'validate-runpath' phase:

https://hydra.gnu.org/build/2078864

The issue is a dangling symlink to "lib/../.index". On 'master', this
symlink is "../.index" which works; this is also the case in the build
directory:

$ ls -al /tmp/guix-build-ruby-ansi-1.5.0.drv-0/ansi-1.5.0/lib/
total 16
drwxr-xr-x 3 marius marius 4096 Jan 16  2015 .
drwxr-xr-x 7 marius marius 4096 Jun  4 15:01 ..
drwxr-xr-x 3 marius marius 4096 Jan 16  2015 ansi
-rw-r--r-- 1 marius marius  426 Jan 16  2015 ansi.rb
lrwxrwxrwx 1 marius marius    9 Jan 16  2015 ansi.yml -> ../.index

So it appars to be somehow mangled during the 'install' phase.

There are a couple of other ruby failures left after the switch to
2.4.0:

https://hydra.gnu.org/eval/109673?compare=master#tabs-now-fail

...but I'm tempted to merge this branch regardless so we can continue
with the 'core-updates' evaluation. Any objections?

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

Re: `ruby-ansi` failure on 'staging'

[Ludovic Courtès]

Hi Marius!

Marius Bakke <mbakke@fastmail.com> skribis:

> On the 'staging' branch (with ruby 2.4), the 'ruby-ansi' package fails
> during the 'validate-runpath' phase:
>
> https://hydra.gnu.org/build/2078864
>
> The issue is a dangling symlink to "lib/../.index". On 'master', this
> symlink is "../.index" which works; this is also the case in the build
> directory:
>
> $ ls -al /tmp/guix-build-ruby-ansi-1.5.0.drv-0/ansi-1.5.0/lib/
> total 16
> drwxr-xr-x 3 marius marius 4096 Jan 16  2015 .
> drwxr-xr-x 7 marius marius 4096 Jun  4 15:01 ..
> drwxr-xr-x 3 marius marius 4096 Jan 16  2015 ansi
> -rw-r--r-- 1 marius marius  426 Jan 16  2015 ansi.rb
> lrwxrwxrwx 1 marius marius    9 Jan 16  2015 ansi.yml -> ../.index
>
> So it appars to be somehow mangled during the 'install' phase.

Weird, is it deterministic?  Or could it have to do with parallel builds
or something?

> There are a couple of other ruby failures left after the switch to
> 2.4.0:
>
> https://hydra.gnu.org/eval/109673?compare=master#tabs-now-fail
>
> ...but I'm tempted to merge this branch regardless so we can continue
> with the 'core-updates' evaluation. Any objections?

Merging sounds reasonable to me.  I’d say go ahead!

Ludo’.