1 |
Chris Gianelloni wrote: |
2 |
|
3 |
>On Mon, 2004-12-27 at 20:36 -0600, William Kilian wrote: |
4 |
> |
5 |
> |
6 |
>>Here are some generic questions I have that reference manual doesn't answer: |
7 |
>> |
8 |
>>Where should I specify ${clst_storedir} and ${clst_sharedir}? |
9 |
>>Environment, spec file, envscript? |
10 |
>> |
11 |
>> |
12 |
> |
13 |
>/etc/catalyst/catalyst.conf and nowhere else. |
14 |
> |
15 |
> |
16 |
OK. It would be nice for those to eventually move to the spec file so |
17 |
that multiple trees of builds can be easily maintained. |
18 |
|
19 |
>>Can the grp/use, tinderbox/use, and livecd/use settings include |
20 |
>>negations (-java) and wildcards? |
21 |
>> |
22 |
>> |
23 |
> |
24 |
>It uses normal portage USE syntax. If portage supports it, then you can |
25 |
>use it. |
26 |
> |
27 |
> |
28 |
Cool. |
29 |
|
30 |
>>For each target, there are possibly use flags, virtuals, keywords, and |
31 |
>>package masks coming possibly from any of: the catalyst spec file; the |
32 |
>>portage_confdir; the snapshot's /usr/portage/profiles; the seed's |
33 |
>> |
34 |
>> |
35 |
> |
36 |
>The portage_confdir would override whatever is in the portage tree, just |
37 |
>like on a normal system. |
38 |
> |
39 |
> |
40 |
>>/usr/portage/profiles; the seed's /etc/make.conf; and the seed's |
41 |
>> |
42 |
>> |
43 |
> |
44 |
>The seed has no /usr/portage/profiles and the make.conf is ignored. |
45 |
> |
46 |
> |
47 |
> |
48 |
>>/etc/portage. For each target, which portage settings come from which |
49 |
>> |
50 |
>> |
51 |
> |
52 |
>Your seed stage should never have anything in /etc/portage, at all. If |
53 |
>you are using a portage_confdir for one target, then you will need it |
54 |
>for them all (unless it doesn't apply). For example, if you make a |
55 |
>change to glibc in your portage_confdir, then you must use the same |
56 |
>portage_confdir throughout the entire build process, or glibc will be |
57 |
>rebuilt in a later target when emerge --newuse is called. Everything |
58 |
>must be consistent across all targets, otherwise the latest target's |
59 |
>environment takes precedence. |
60 |
> |
61 |
> |
62 |
> |
63 |
>>sources? A table for each target that illustrates what settings comes |
64 |
>>from what location would provide me with a lot more clarity. |
65 |
>> |
66 |
>> |
67 |
> |
68 |
>It works identical to portage itself. |
69 |
> |
70 |
> |
71 |
> |
72 |
That explanation helps a lot. Thanks! Let me see if I have this straight: |
73 |
|
74 |
The seed is basically devoid of portage configuration data because it |
75 |
shouldn't have an /etc/portage or a /usr/portage/profiles and its |
76 |
make.conf (if any) is ignored. The spec file use flags during an |
77 |
applicable build are functionally equivalent to USE= in make.conf on a |
78 |
normal system. The profile information is provided by the snapshot and |
79 |
portage_confdir provides the information a normal system's /etc/portage |
80 |
directory would provide. The snapshot's profile, target/use, and |
81 |
portage_confdir are used the same way as portage on a normal system uses |
82 |
its profile, USE in make.conf, and portage_confdir. |
83 |
|
84 |
Let me know if I'm misunderstanding something. |
85 |
|
86 |
Cheers, |
87 |
|
88 |
William Kilian |
89 |
|
90 |
-- |
91 |
gentoo-catalyst@g.o mailing list |