1 |
I don't think I would have realized this implementation myself. Is this |
2 |
xpak piggybacking documented somewhere?Can you refer me? |
3 |
BTW: |
4 |
amit0 Installation # man xpak |
5 |
No manual entry for xpak |
6 |
amit0 Installation # man 5 xpak |
7 |
No entry for xpak in section 5 of the manual |
8 |
amit0 Installation # eix -S xpak |
9 |
No matches found. |
10 |
|
11 |
Amit |
12 |
|
13 |
Zac Medico wrote: |
14 |
> Amit Dor-Shifer wrote: |
15 |
> > When you say ut checks CHOST/keywording, where are those definitions |
16 |
> > stored for the binary pkg? |
17 |
> |
18 |
> It's appended to the tail end of the tbz2, in xpak format (see `man |
19 |
> 5 xpak`). |
20 |
> |
21 |
> > I see one instance of CHOST in the 'Packages' index on the BINHOST. Is |
22 |
> > that the variable emerge is comparing against? If not, where is it? the |
23 |
> > tbz itself holds just the binaries. |
24 |
> |
25 |
> Yes, for remote packages, the Package index contains equivalent data |
26 |
> to the actual xpak segments from the remote packages. The CHOST for |
27 |
> individual packages is only shown in cases when it differs from the |
28 |
> CHOST in the header of the Packages file. |
29 |
> |
30 |
> > Manually modifying a/m CHOST to 'ppc' didn't stop emerge from |
31 |
> > successfully merging a package on an amd64 target (I've removed |
32 |
> > /usr/portage/packages from target before emerging). |
33 |
> |
34 |
> The CHOST from your local configuration (typically from make.conf or |
35 |
> inherited from your profile) is compared to the CHOST of the binary |
36 |
> package. You can also use ACCEPT_CHOSTS if you want to accept more |
37 |
> than one CHOST (see `man 5 make.conf`). |