1 |
Hemmann, Volker Armin posted |
2 |
<200511162032.10112.volker.armin.hemmann@××××××××××××.de>, excerpted |
3 |
below, on Wed, 16 Nov 2005 20:32:10 +0100: |
4 |
|
5 |
> On Wednesday 16 November 2005 18:30, Duncan wrote: |
6 |
>> Hemmann, Volker Armin posted |
7 |
>> <200511161606.07866.volker.armin.hemmann@××××××××××××.de>, excerpted |
8 |
>> |
9 |
>> below, on Wed, 16 Nov 2005 16:06:07 +0100: |
10 |
>> > So I would suspect some miscompilation by gcc4 ... there are some |
11 |
>> > reasons, that it is still hardmasked ;) |
12 |
>> |
13 |
>> Yes, but usually it's simply that it's stricter and therefore refuses to |
14 |
>> compile whatever. Additionally, KDE 3.5 is supposed to be gcc-4.x tested |
15 |
>> upstream, as they are doing the gcc-4.x fvisibility stuff that had to be |
16 |
>> pulled from kde 3.4.whatever. |
17 |
> |
18 |
> hm, very intersting, but the gcc is the only difference. Hm, or is it the |
19 |
> binutils? |
20 |
> I have |
21 |
> sys-devel/binutils |
22 |
> Latest version available: 2.16.1 |
23 |
> Latest version installed: 2.16.1 |
24 |
|
25 |
sys-devel/binutils |
26 |
Latest version available: 2.16.91.0.1 |
27 |
Latest version installed: 2.16.91.0.1 |
28 |
|
29 |
I'm not sure if it's still hard-masked or if it is now ~arch, but it was |
30 |
hard-masked when I overlayed and keyworded it, as it was a dependency of |
31 |
the (also still hardmasked) gcc-4.x. That's what I met when I said |
32 |
"related" binutils -- I had to unmask both (tho I had a choice of possible |
33 |
binutils versions to unmask and merge) to get gcc4. |
34 |
|
35 |
>> Everything else seems to be working great, including one thing I had a |
36 |
>> problem with in beta2! =8^) |
37 |
>> |
38 |
> |
39 |
> hm, for me, kde 3.5-beta2 and rc1 are even more stable than any other release |
40 |
> ever. |
41 |
> |
42 |
> I had not one crash or misbehaviour so far - which is very promising. |
43 |
|
44 |
The beta2 thing wasn't major. It was xmms-kde, a kicker applet to control |
45 |
xmms/noatun/xmpeg. It would launch sometimes, but not others, complaining |
46 |
about a library it couldn't find. Actually, now that I think about it, |
47 |
the problem and why it now works was probably that I needed to run |
48 |
fix_libtool_files.sh, and did so after the -rc1 merge died shortly after |
49 |
initial run. After running fix_libtool_files.sh, the merge finished. |
50 |
xmms-kde probably had a dependency on something configured with an la file |
51 |
that pointed to a stale gcc version, so running the script fixed it as |
52 |
well. |
53 |
|
54 |
> But if you want to recompile with ggcc3.4 don't you need to recompile qt and |
55 |
> kdelibs too? |
56 |
> I am not sure, but prior experiences with 3.2 and 3.3 told me, that |
57 |
> kdepackages can be pretty annoying if not compiled with the exact same |
58 |
> compiler. |
59 |
|
60 |
Well, the major problem, is libstdc++. The gcc-4.0 versions are supposed |
61 |
to be generally backward compatible with the 3.4 versions, and I've found |
62 |
it to be generally so. There are occasional issues, but they are usually |
63 |
fixable if one can find the correct combination of 3.4 and 4.0 compiled |
64 |
applications and libraries. Running fix_libtool_files.sh and recompiling |
65 |
the target application often helps, as was apparently the case with |
66 |
xmms-kde, but to do that, you have to know which specific version is |
67 |
causing the error, so you can run it against that version. This of course |
68 |
isn't a big issue with binary distributions where everything in a release |
69 |
is compiled using the same gcc, thus against the same libstdc++. |
70 |
|
71 |
You are correct, however, it can be "annoying", at times. |
72 |
|
73 |
FWIW, I'm now running a gcc-3.4.4 compiled kmail, and no problems yet, but |
74 |
I've only had a few spams come thru since I restarted it, so I can't say |
75 |
the problem's definitely fixed. So far tho, so good (said the guys |
76 |
jumping out of the twin towers as they passed the 10th floor, hopefully |
77 |
this experience ends rather better, and hopefully noone finds this in |
78 |
/too/ bad a taste). |
79 |
|
80 |
-- |
81 |
Duncan - List replies preferred. No HTML msgs. |
82 |
"Every nonfree program has a lord, a master -- |
83 |
and if you use the program, he is your master." Richard Stallman in |
84 |
http://www.linuxdevcenter.com/pub/a/linux/2004/12/22/rms_interview.html |
85 |
|
86 |
|
87 |
-- |
88 |
gentoo-amd64@g.o mailing list |