1 |
On Wed, 16 May 2007 17:47:37 +0200, Enrico Weigelt wrote: |
2 |
|
3 |
>> Lots? If you'd posted this yesterday, I would have been able to recall |
4 |
> > the last time I was hit with one. |
5 |
> |
6 |
> At least several. I didn't find an good solution for checking |
7 |
> the whole tree yet, so I yet know some. Good candidates are |
8 |
> where PDEPENDs occour. For example the Xserver. |
9 |
|
10 |
As Bo has already explained, PDEPENDS cannot cause circular dependencies, |
11 |
in fact they prevent them. |
12 |
|
13 |
> > I did get a circular dependency today, sdl and directfb and guess |
14 |
> > what? The error message also contained the solution, which was to |
15 |
> > temporarily change a USE flag. |
16 |
> |
17 |
> *rofl* what a good solution. really clean. gread idea. |
18 |
|
19 |
Yes, it is a good solution, because the circular dependency is caused by |
20 |
changed USE flags in the first place. You did notice my use of the word |
21 |
"temporarily", didn't you? If A depends on B and B depends on A, you |
22 |
build A without support for B, then you can safely install B and A again |
23 |
with the features you wanted. This is nothing to do with portage but a |
24 |
consequence of building packages from different projects from source. |
25 |
Naturally, you won't see this with a binary distro, because someone has |
26 |
already gone through the "build A without B support, build B, build A |
27 |
again" process, but without the benefit of USE flags to make it much |
28 |
easier. |
29 |
|
30 |
Having said that, portage ought to be able to handle this situation |
31 |
better than it does, and work in continuing in that direction. Anal |
32 |
sarcasm does not shorten the development cycle for such features. |
33 |
|
34 |
> > The great thing about free software is that it all comes with a full |
35 |
> > money back guarantee. So before you start shooting your mouth off when |
36 |
> > something produced by volunteers in their own time fails to work for |
37 |
> > you, stop and remember how much you paid for it and why. |
38 |
> |
39 |
> You don't need to give me lesson on OSS, I've did enough work OSS |
40 |
> projects over last 12 years to know how it works. |
41 |
|
42 |
Really, that comes as something of a surprise after reading your posts. |
43 |
|
44 |
> And I'm really tired of trying to talk with learn-resistent devs who |
45 |
> clearly expressed that they don't any of my help and told be to stay |
46 |
> away from b.g.o. So I'm maintaining my own overlay and concentrate on |
47 |
> getting things running instead of wasting time with certain devs. |
48 |
|
49 |
"Certain devs"? Your previous batch of insults was aimed at "the Gentoo |
50 |
devs". Are you becoming more selective over whom you wish to impugn? |
51 |
|
52 |
> > Since you are clearly able to solve a problem that comes fro upstream |
53 |
> > and which the combined might of SUSE and Gentoo has failed to do, |
54 |
> |
55 |
> Which problem from the upstream ? |
56 |
|
57 |
Interdependency of unrelated packages. |
58 |
|
59 |
> The problem w/ x11-base/xorg-server are the PDEPENDs on (external) |
60 |
> driver packages. I dont see any valid reason for depending the |
61 |
> Xserver on drivers, which themselfes depend on the Xserver. |
62 |
|
63 |
The server does not depend on the drivers, it depends on them, it |
64 |
wouldn't be much use without at least three drivers. |
65 |
|
66 |
> IMHO, there were days where it had been done so (when PDEPEND did not |
67 |
> yet exist). I don't know why this had changed, probably just to get |
68 |
> an new feature widely used. (BTW: I do not see any valid reason for |
69 |
> PDEPEND anyways) |
70 |
|
71 |
Which probably says more about your understanding of how things work than |
72 |
it does about their usefulness. |
73 |
|
74 |
> In fact, I'll provide an solution. It will be published within the |
75 |
> OSS-QM project. But I won't waste any second on filing any bug, |
76 |
> just for that it's marked invalid by folks like Jakub. |
77 |
|
78 |
Jakub is no longer a bug-wrangler, or a dev, he retired last month. |
79 |
|
80 |
|
81 |
-- |
82 |
Neil Bothwick |
83 |
|
84 |
New: Different color from previous model. |