1 |
On 5/27/06, Alexander Skwar <listen@×××××××××××××××.name> wrote: |
2 |
> > Stop using ~arch packages, or stop whining. |
3 |
> |
4 |
> No, I won't do neither. The GWN and the upgrade doc used to say, |
5 |
> that an upgrade is (basically) riskless. |
6 |
|
7 |
Well I can't force you to do anything. You found a problem, reported |
8 |
a bug, and got the documentation fixed. Great, all useful |
9 |
contributions to a community supported distribution. |
10 |
|
11 |
However you have also attacked the Gentoo devs. You have insinuated |
12 |
that they are lazy and/or careless. You do remember that Gentoo devs |
13 |
are unpaid volunteers doing this in their spare time, right? And this |
14 |
is how you choose to thank them for that gift? |
15 |
|
16 |
> That's wrong - as it has been confirmed and corrected now. If that |
17 |
> warning would have been there in the gcc-upgrade doc, everything |
18 |
> would have been fine. |
19 |
|
20 |
And it will be when gcc 4.1.1 goes out of ~arch, so the stable users |
21 |
will know they need to do an emerge -e world. Yeah for them! |
22 |
|
23 |
> Bullshit. |
24 |
|
25 |
My thoughts exactly... |
26 |
|
27 |
> I'd expect them to do testing and not give so bold statements |
28 |
> as "The upgrade should be incredibly easy and require no additional |
29 |
> work to install and use. " without making VERY sure, that this |
30 |
> is actually true. |
31 |
|
32 |
They added it to ~arch to make VERY sure that this was true, and it wasn't. |
33 |
|
34 |
> And what's also irritating are so many small errors, like files |
35 |
> with non-matching filesizes/checksums in the digests. |
36 |
|
37 |
The filesize issue is probably because the distfile changed without an |
38 |
change in the ebuild version. So you can get a new Manifest (from |
39 |
emerge --sync) that doesn't match the actual distfile you have, and |
40 |
get a filesize mismatch the next time you try to merge that same |
41 |
version, e.g. when doing an emerge -e world. This is rare, but it |
42 |
does happen. |
43 |
|
44 |
The evidence that this is in any way related to the gcc upgrade is |
45 |
pretty thin... |
46 |
|
47 |
> > I just upgraded to gcc-4.1 and pruned 3.4.6, and KDE, koffice, OOo, |
48 |
> > and mozilla all still load and run fine. |
49 |
> |
50 |
> Did you yet re-compile Qt 3 and Qt 4? No? |
51 |
|
52 |
For the sake of argument, I just did. Guess what? The only "bad" |
53 |
thing that happened is my KDE theme went away. No big deal, I've seen |
54 |
it before when upgrading qt, and although I'm not sure what causes it, |
55 |
I know that remerging kdelibs fixes it. |
56 |
|
57 |
But of course this is meaningless...the fact that some people have no |
58 |
problems doesn't change the fact that some do. |
59 |
|
60 |
> Then you're experiences just don't count. KDE broke on my |
61 |
> system, when I recompiled Qt. Before the recompile, KDE was fine. |
62 |
> As I've wrote in lengths on the bug report. Seems you've not read |
63 |
> it - why not? Why am I writing reports and *also* post links |
64 |
> here? |
65 |
|
66 |
I *did*. And it was very good and useful report, even though it was a |
67 |
duplicate of another report. |
68 |
|
69 |
Look, I am not claiming there the gcc upgrade is seamless and easy. I |
70 |
am also *not* suggesting that the fact that it works on my system is |
71 |
proof that there are no problems. But it does mean that the upgrade |
72 |
can *appear* to be seamless and easy, and only fail in specific cases |
73 |
on specific systems. |
74 |
|
75 |
> Did you try to compile glib? No? Then I guess you've done no testing. |
76 |
|
77 |
Um, I'm a ~arch user, so I am *always* testing. But no, I didn't |
78 |
recompile just glib to see if it would work in a mixed environment. |
79 |
Nor did I re-compile each of the other 835 packages installed on my |
80 |
system in turn with 4.1.1 to see if they would work in a mixed |
81 |
environment. |
82 |
|
83 |
But, did you happen to notice that the original poster of that glib |
84 |
bug wasn't even using gcc 4.1? How did you decide that it was a gcc |
85 |
upgrade problem? Indeed it looks more like a libtool issue to me... |
86 |
|
87 |
Do you have some plan for testing new gcc releases in a mixed |
88 |
compilation environment that will guarantee no compatibility problems? |
89 |
Something that won't take a year to complete so that new gcc versions |
90 |
can move out of p.mask? |
91 |
|
92 |
IMO, it would be much simpler and better to tell people to always do |
93 |
an emerge -e world when upgrading gcc... |
94 |
|
95 |
-Richard |
96 |
|
97 |
-- |
98 |
gentoo-user@g.o mailing list |