[Slackbuilds-users] docker build failure

[mancha]

On Sat, 08 Feb 2014 20:07:21 +0000 "Ivan Stefanic" wrote:
>I am running Slackware64 14.1. I also had problems building google-
>go-lang, 
>same as Antonio Hernández Blas in another thread, but commenting 
>out 
>'set -e' made it build. I don't know if this caused problems with 
>docker 
>but that's how my build differs from what it should be, and google-
>go-lang is 
>a build-time dependency for docker. 
>
>docker.Slackbuild output: 
>
>chmod: cannot operate on dangling symlink 
>'./utils/testdata/fs/a/d'
>chmod: cannot operate on dangling symlink 
>'./utils/testdata/fs/a/f'
>patching file graphdriver/btrfs/btrfs.go
># WARNING! I don't seem to be running in the Docker container.
># The result of this command might be an incorrect build, and will 
>not be
>#   officially supported.
>#
># Try this instead: make all
>#
>
>---> Making bundle: dynbinary (in bundles/0.8.0/dynbinary)
>Created binary: /tmp/SBo/docker-
>0.8.0/bundles/0.8.0/dynbinary/dockerinit-0.8.0
># github.com/dotcloud/docker/graphdriver/btrfs
>../package-
>docker/usr/share/gocode/src/github.com/dotcloud/docker/graphdriver/
>btrfs/btrfs.go:8:25: fatal error: btrfs/ioctl.h: No such file or 
>directory
> #include <btrfs/ioctl.h>
>                         ^
>compilation terminated.
>
>End output. 
>
>Any ideas?
>
>-- 
>Ivan Stefanic <stefanicivan at gmail.com>

Hi Vincent.

On x86_64 the docker build process uses btrfs.go from $PKG
but that copy is made prior to applying the btrfs reversion
fix - triggering Ivan's build issue above.

I've re-ordered things in docker.Slackbuild so source files
get copied to $PKG post-patch (I think we want that anyways,
regardless of build issues, right?).

After this change it builds without errors on x86_64 but I 
don't use docker so I can't confirm what gets built works.

Willy didn't have these issues when building on 32-bit x86
so it seems there it uses btrfs.go from /tmp/SBo/docker-0.8.0.
Why 32-bit and 64-bit builds differ? No clue. Ideas?

--mancha