1 |
On Tue, 17 Apr 2012 18:41:55 +0300 |
2 |
Nikos Chantziaras <realnc@×××××.com> wrote: |
3 |
|
4 |
> On 17/04/12 18:34, Nikos Chantziaras wrote: |
5 |
> > On 17/04/12 18:10, Alan McKinnon wrote: |
6 |
> >> On Tue, 17 Apr 2012 18:01:43 +0300 |
7 |
> >> Nikos Chantziaras<realnc@×××××.com> wrote: |
8 |
> >> |
9 |
> >>> emerge --depclean seems to have bugged out for me: |
10 |
> >>> |
11 |
> >>> * Dependencies could not be completely resolved due to |
12 |
> >>> * the following required packages not being installed: |
13 |
> >>> * |
14 |
> >>> * =dev-libs/openssl-0.9.8* pulled in by: |
15 |
> >>> * app-emulation/vmware-workstation-8.0.2.591240 |
16 |
> >>> * |
17 |
> >>> * dev-libs/openssl:0.9.8 pulled in by: |
18 |
> >>> * www-client/google-chrome-19.0.1084.24_beta131971 |
19 |
> >>> |
20 |
> >>> It says to do a "emerge --update --newuse --deep --with-bdeps=y |
21 |
> >>> @world". Which is what I just did to begin with (and running it |
22 |
> >>> again doesn't result in anything getting emerged.) |
23 |
> >>> |
24 |
> >>> It also says: |
25 |
> >>> |
26 |
> >>> * Also, note that it may be necessary to manually uninstall |
27 |
> >>> * packages that no longer exist in the portage tree, since it may |
28 |
> >>> * not be possible to satisfy their dependencies. |
29 |
> >>> |
30 |
> >>> However, dev-libs/openssl is installed and seems to be in the |
31 |
> >>> tree: |
32 |
> >>> |
33 |
> >>> $ eix -e dev-libs/openssl |
34 |
> >>> [I] dev-libs/openssl |
35 |
> >>> Available versions: |
36 |
> >>> (0.9.8) 0.9.8r (~)0.9.8s 0.9.8s-r1 0.9.8t 0.9.8u |
37 |
> >>> (0) 1.0.0d 1.0.0e (~)1.0.0e-r1 (~)1.0.0f 1.0.0f-r1 |
38 |
> >>> 1.0.0g 1.0.0h **1.0.1 |
39 |
> >>> {{bindist gmp kerberos rfc3779 sse2 static-libs test vanilla |
40 |
> >>> zlib}} Installed versions: 0.9.8u(0.9.8)(22:32:12 12/03/12)(sse2 |
41 |
> >>> zlib -bindist -gmp -kerberos -test) 1.0.0h(22:33:20 12/03/12)(sse2 |
42 |
> >>> zlib -bindist -gmp -kerberos -rfc3779 -static-libs -test) |
43 |
> >>> |
44 |
> >>> |
45 |
> >> |
46 |
> >> I think you should file a bug against www-client/google-chrome |
47 |
> >> |
48 |
> >> It wants the exact version dev-libs/openssl:0.9.8 but the oldest in |
49 |
> >> portage is dev-libs/openssl:0.9.8r |
50 |
> >> |
51 |
> >> Chances are the dev simply left off the version suffix by accident |
52 |
> > |
53 |
> > That can't be it. There's a wildcard at the end: |
54 |
> > |
55 |
> > =dev-libs/openssl-0.9.8* |
56 |
> > |
57 |
> > which should match 0.9.8u. |
58 |
> |
59 |
> No wait, that's vmware. The chrome ebuild pulls a slot: |
60 |
> |
61 |
> dev-libs/openssl:0.9.8 |
62 |
> |
63 |
> This also matches correctly. The 0.9.8 slot matches all those |
64 |
> versions. So no problem with that either. |
65 |
|
66 |
I missed that colon for the slot. |
67 |
So this certainly looks buggy on the part of depclean. Now to find out |
68 |
what's going on. |
69 |
|
70 |
One thing that strikes me as a possibility is that the highest numbered |
71 |
version has the lower slot number (0). I know that SLOT names are |
72 |
just names and not supposed to be treated like they form an order, but |
73 |
I've seen odd things like this before. I think KDE did it to me once. |
74 |
|
75 |
What happens if you emerge openssl:0.9.8, let it go into world, then |
76 |
run depclean? The build takes 2 minutes and you can easily remove it |
77 |
from world later. |
78 |
|
79 |
-- |
80 |
Alan McKinnnon |
81 |
alan.mckinnon@×××××.com |