1 |
Yannick Koehler wrote: |
2 |
> -----BEGIN PGP SIGNED MESSAGE----- |
3 |
> Hash: SHA1 |
4 |
> |
5 |
> |
6 |
>>Appart from the trust issue, the major problem is comparing the two |
7 |
>>systems. It is unlikely they are exactly the same. But even if they |
8 |
are, it |
9 |
>>is a hell of a job finding out. The only way such a distribution for |
10 |
>>binaries works is with a binary only distribution like,..... (you know |
11 |
>>who). |
12 |
> |
13 |
> |
14 |
> Portage keeps certain file which help on this such as |
15 |
> /var/db/pkg/<cat>/<name>/CFLAGS and USE. Maybe they could be named |
16 |
on the |
17 |
> ftp server to |
18 |
> |
19 |
> <name>.specification |
20 |
> <name>.tar.gz |
21 |
> |
22 |
> So that emerge download the .specification, validate the similarity |
23 |
and then |
24 |
> propose or take it from the distribution system. Basically the same as |
25 |
> ccache ;-) |
26 |
> |
27 |
|
28 |
I like the idea. I was thinking of something similar. |
29 |
I think it's possible to hash the use flags used to build |
30 |
the package and compare it to the package to be downloaded. |
31 |
|
32 |
However I doubt of the practical usefulness of a global peer-to-peer |
33 |
solution. I have an 128bps upstream bandwidth and everyone going to |
34 |
copy compiled kde-3.0 from me would compile it faster on PIII500 (~). |
35 |
Since the userbase not so big as *pster and the number of combinations |
36 |
of use flags is big, it is not very likely to find a package provided by |
37 |
a fast host. |
38 |
|
39 |
I think it can be useful in a lan where, for whatever reasion, the |
40 |
machines doesn't share the same use flag configuration. If a package |
41 |
desn't use the "mysql" use-flag then it doesn't depend of having it or not. |
42 |
|
43 |
Marko |