1 |
On 11/09/2013 22:46, Timur Aydin wrote: |
2 |
> On 09/11/13 23:13, Alan McKinnon wrote: |
3 |
>> I can't reproduce that fault here, and google hits on the matter seem to |
4 |
>> point towards stale metadata referencing eclasses that no longer exist. |
5 |
>> I have a hunch you do not have valid metadata for your local overlay, so |
6 |
>> I suggest: |
7 |
>> |
8 |
>> 1. delete the eclass directory from your overlay, run "emerge |
9 |
>> --metadata" and emerge samba again. See what that does. |
10 |
> |
11 |
> I have tried this, and the same problem happened, where emerge |
12 |
> complained that pam.eclass wasn't found. |
13 |
> |
14 |
>> 2. Set PORTDIR_CACHE_METHOD and/or OVERLAY_CACHE_METHOD explicitly in |
15 |
>> make.conf, the best reference for these is in the eix man page |
16 |
> |
17 |
> While reading up on the various methods in the eix man page, I started |
18 |
> to wonder whether a setting in /usr/local/portage/metadata/layout.conf |
19 |
> was causing this problem. When testing, layout.conf had only a single |
20 |
> entry, which was: |
21 |
> |
22 |
> masters = |
23 |
> |
24 |
> I remembered an error message from portage a while ago, which suggested |
25 |
> setting this to "gentoo" for backward compatibility. I tried that, and |
26 |
> the problem went away :) I was able to emerge samba from the overlay |
27 |
> successfully. Does this make sense? |
28 |
|
29 |
|
30 |
|
31 |
/headbang :-) |
32 |
|
33 |
Yes of course, it make a great deal of sense now. Basically, your local |
34 |
overlay had no idea where the parent portage tree is or how to find it |
35 |
so couldn't find the eclass directory. |
36 |
|
37 |
As I understand it, this information used to be hard-coded magic and |
38 |
overlays would "just know where to look". Since recently, you have to |
39 |
configure it explicitly and not use hidden super-magic. |
40 |
|
41 |
You would have been getting confusing messages in emerge output about a |
42 |
faulty masters setting for overlays, pity we didn't spot that up front. |
43 |
Double pity that there wasn't a clear message or news item about what |
44 |
the error meant and the impact.... |
45 |
|
46 |
|
47 |
-- |
48 |
Alan McKinnon |
49 |
alan.mckinnon@×××××.com |