1 |
On Thursday 01 July 2010 00:15:34 Mick wrote: |
2 |
> > I remember now what I eventually did to fix all my libpng problems back |
3 |
> > |
4 |
> > then: |
5 |
> > |
6 |
> > |
7 |
> > unmerge libpng |
8 |
> > delete everything left with libpng in it's name |
9 |
> > emerge -pvuND world just to see what was now busted |
10 |
> > same with revdep-rebuild |
11 |
> > re-emerge libpng and let portage pick the version to merge |
12 |
> > emerge -avuND world |
13 |
> > revdep-rebuild |
14 |
> > |
15 |
> > repeat till no problems reported |
16 |
> > |
17 |
> > |
18 |
> > and lafilefixer --justfixit somewhere in the mix as well just for good |
19 |
> > measure. |
20 |
> > |
21 |
> > |
22 |
> > |
23 |
> > This is called the "throw shit at the wall and hope some of it sticks" |
24 |
> > method of updating systems. By happy good fortune, it worked out for me. |
25 |
> > |
26 |
> > |
27 |
> > |
28 |
> > aka the dark underside of source-based systems :-) |
29 |
> |
30 |
> I'm sure that I followed your scientific approach too because my system |
31 |
> still works ... although I may have lost the will to live somewhere in |
32 |
> between the lafilefixer and revdep-rebuild for the nth time and decided to |
33 |
> emerge -e world. :-)) |
34 |
> |
35 |
> Strangely enough two other x86 boxen were updated without any much drama. |
36 |
|
37 |
:-) |
38 |
|
39 |
On the front page of gentoo.org several blogs are reference that mention this |
40 |
issue. In one of them, the author shows evidence that the same binary was |
41 |
trying to link at one point to a 32 bit lib and at another point link to a 64 |
42 |
bit lib..... All random of course, causing no end of mysterious failures, and |
43 |
explains why some people had success re-merging cairo, other with something |
44 |
else .... |
45 |
|
46 |
Truly bizarre. But it might go a ways to explaining why your 32 bit systems |
47 |
were unaffected? |
48 |
|
49 |
-- |
50 |
alan dot mckinnon at gmail dot com |