1 |
On 11/13/18 10:50 AM, Zac Medico wrote: |
2 |
> On 11/11/18 12:53 PM, Michał Górny wrote: |
3 |
>> Hi, |
4 |
>> |
5 |
>> Ok, here's the second version integrating the feedback received. |
6 |
>> The format is much simpler, based on nested tarballs inspired by Debian. |
7 |
>> |
8 |
>> The outer tarball is uncompressed and uses '.gpkg.tar' suffix. It |
9 |
>> contains (preferably in order but PM should also handle packages with |
10 |
>> mismatched order): |
11 |
>> |
12 |
>> 1. Optional (but recommended) "gpkg: ${PF}" package label that can be |
13 |
>> used to quickly distinguish Gentoo binpkgs from regular tarballs |
14 |
>> (for file(1)). |
15 |
>> |
16 |
>> 2. "metadata.tar${comp}" tarball containing binary package metadata |
17 |
>> as files. |
18 |
>> |
19 |
>> 3. Optional "metadata.tar${comp}.sig" containing detached signature |
20 |
>> for the metadata archive. |
21 |
>> |
22 |
>> 4. "contents.tar${comp}" tarball containing files to be installed. |
23 |
>> |
24 |
>> 5. Optional "contents.tar${comp}.sig" containing detached signature for |
25 |
>> the contents archive. |
26 |
> |
27 |
> We need to establish the procedure for signature verification of the |
28 |
> files in "contents.tar${comp}" at any point in the future *after* they |
29 |
> have been installed. In order to identify corruption of a particular |
30 |
> installed file, we'll need separate digests for each of the installed |
31 |
> files, and a signature covering the separate digests. |
32 |
|
33 |
We need separate digests for the files in "metadata.tar${comp}" too, for |
34 |
the same reason. Note the environment.bz2 is mutable because it is |
35 |
deserialized/reserialized for each pkg_* phase. If the installation |
36 |
process has access to a trusted signing key, it can sign environment.bz2 |
37 |
after each mutation. |
38 |
-- |
39 |
Thanks, |
40 |
Zac |