Gentoo Archives: gentoo-science

From: Steven Trogdon <strogdon@×××××.edu>
To: gentoo-science@l.g.o
Subject: Re: [gentoo-science] [sage-on-gentoo] polybori && amd64
Date: Fri, 20 Aug 2010 14:12:35
Message-Id: 1282313417.19920.0@pavilion64
In Reply to: Re: [gentoo-science] [sage-on-gentoo] polybori && amd64 by Christopher Schwan
1 Hi Christopher,
2
3 On 08/20/2010 03:06:53 AM, Christopher Schwan wrote:
4 > Hi Steve,
5 >
6 > On Thursday 19 August 2010 19:32:23 Steven Trogdon wrote:
7 >> On 08/19/2010 03:04:54 AM, Christopher Schwan wrote:
8 >> > Hi,
9 >> >
10 >> > On Thursday 19 August 2010 08:14:08 Steven Trogdon wrote:
11 >> >> Hello,
12 >> >>
13 >> >> Both polybori-0.6.4-r1 and polybori-0.6.5 fail to build here on amd64
14 >> >> for the same reason. The important snippet is:
15 >> >>
16 >> >> rm: cannot remove
17 >> >> `/var/tmp/portage/sci-mathematics/polybori-0.6.4-r1/image//usr/lib64/lib
18 >> >> *.s o*': No such file or directory
19 >> >>
20 >> >> * ERROR: sci-mathematics/polybori-0.6.4-r1 failed:
21 >> >> * failed to remove shared objects
22 >> >> *
23 >> >> * Call stack:
24 >> >> * ebuild.sh, line 54: Called src_install
25 >> >> * environment, line 5124: Called die
26 >> >> * The specific snippet of code:
27 >> >> * rm "${ED}"/usr/$(get_libdir)/lib*.so* || die "failed to remove
28 >> >>
29 >> >> shared objects"
30 >> >>
31 >> >> *
32 >> >>
33 >> >> Under the /image directory polybori installs the libs under /usr/lib
34 >> >> instead of /usr/lib64. This seemed to be an issue previously with
35 >> >> polybori.
36 >> >
37 >> > Fixed in overlay - I will try to find find out how to install to lib64 on
38 >> > amd64.
39 >>
40 >> With the present commit I get multilib-strict failures. I forgot about
41 >> this! I put the following:
42 >>
43 >> sed -i "s:DevelInstPath('lib'):DevelInstPath('$(get_libdir)'):g" \
44 >> SConstruct || die "sed failed"
45 >>
46 >> in src_prepare and libs are installed under /usr/lib64 on amd64 and under
47 >> /usr/lib on my 32bit chroot. I'm not sure if this is the "correct" way to
48 >> fix things but maybe it'll give some direction.
49 >
50 > Thank you for your info! I will update polybori later this day (I currently
51 > do
52 > not have access to an amd64 machine) and see if its possible to fix this
53 > issue
54 > with an command line argument (that would be the correct way).
55 >
56 > Besides polybori, do you have any other multilib-strict failures for sage-
57 > related ebuilds ?
58 >
59
60 I haven't seen other ebuilds recently with this problem. I looked at adjusting
61 the PARAMS variable but could find nothing to pass to scons that would do the
62 job. I could have missed something.
63
64 >>
65 >> >> Steve
66 >> >
67 >> > Christopher
68 >>
69 >> Steve
70 >
71 > Christopher
72 >
73
74 Steve