guix-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Test failure in util-linux


From: Ludovic Courtès
Subject: Re: Test failure in util-linux
Date: Fri, 10 Apr 2015 15:09:54 +0200
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/24.4 (gnu/linux)

Hello,

address@hidden (Carlos Sánchez de La Lama) skribis:

> I am trying to "guix system build" a smallish configuration (most
> settings at their default values). I have installed guix on a freshly
> new i686 gNewSense 3.1 (which uses kernel 2.6.32).
>
> The system build fails during "check" phase of util-linux-2.25.2. I have
> dug a little on and found the error comes from the fscanf in
> util-linux-2.25.2/sys-utils/ipcutils.c:113
> expecting 16 values in /proc/sysvipc/shm, while min having only 14. My
> impression is util-linux checks have a dependency on the kernel being
> fairly recent.

Oh, OK.  We could work around it, but since that requires a full
rebuild, we’ll have to schedule it for the next core update cycle.

That said, 2.6.32 is really old, so we probably don’t want to invest too
much in that.

> My points:
>
> 1) should check phase be disabled for util-linux depending on some
> kernel version check?

Yes, we could patch this specific test or code snippet.

> 2) I tried adding "#:tests? #f" to my own modified ~/guix/linux.scm
> (copied from system-wide
> /usr/local/share/guile/site/2.0/gnu/packages/linux.scm).
> This allowed building with
> "guix package -L $HOME/guix build util-linux"
> but system build still fails during the tests (would seem guix system
> ignores -L flag and takes system-wide recipe).

Right.  “guix build” now sees your ‘util-linux’ package, but the whole
package DAG uses explicit references to the <package> objects, and so
yours is ignored.

You may be able to get around it by defining your own (gnu packages
linux) package.  The downside is that it would have to basically be a
copy of the original one with just #:tests? #f added.

If Guile had a #:use-module-next form (akin to #include_next in C), we
could do something better here.

I hope this makes sense!

Ludo’.



reply via email to

[Prev in Thread] Current Thread [Next in Thread]