1 |
Patrick Lauer <patrick@g.o> wrote: |
2 |
> |
3 |
> Without binpkg support I'd feel the need to hack it up, just to get things |
4 |
> fast enough. |
5 |
|
6 |
binpkg support has some severe problems currently, anyway. |
7 |
I already described it in the bug, but since perhaps the |
8 |
description was not clear, I repeat it here: |
9 |
|
10 |
I regularly build binpackages for some client. |
11 |
This client is updated only very infrequently |
12 |
(once every few months). |
13 |
|
14 |
Of course, before I emerge the binpkgs, I update the |
15 |
portage tree on that client, and also copy my |
16 |
/etc/portage/package* directories. |
17 |
|
18 |
Then it turns out that I can use *less* than half of |
19 |
my binpkgs on the client: The binpkgs are just ignored. |
20 |
|
21 |
One reason I found: I had temporarily set ~ARCH for some |
22 |
package in /etc/portage/package.accept_keywords |
23 |
when building the package, and removed the entry again |
24 |
(and thus also on the client) once the package has become |
25 |
stable. I hope that you agree that my expectation is |
26 |
sane that the binpkg should be used anyway - but it isn't. |
27 |
|
28 |
For others I do not know the reason, but probably it is |
29 |
very similar that some variables/settings at the time |
30 |
of building the package were different from the current |
31 |
ones in the tree. |
32 |
|
33 |
So, I do not know all the details, but currently |
34 |
binpkg support is somewhat broken already. |
35 |
What we would need is *more* pushing of information of |
36 |
the current tree data to binpkgs, not *less*. |
37 |
So dynamic deps (or another update mechanism) |
38 |
appear to be a first step in *fixing* support for binpkgs... |