1 |
Akos Maroy wrote: |
2 |
|
3 |
> we already have portage working, with an old snapshot of the portage |
4 |
> tree (arm keywords still included). see |
5 |
> http://gentooforzaurus.opensistemas.com/ |
6 |
> |
7 |
> the idea at the moment is to mount gentoo-related files through nfs, as |
8 |
> it is quite large, and does not fit on most memory cards for the Zaurus. |
9 |
> |
10 |
|
11 |
I don't doubt this. The smallest drive I managed to kludge gentoo on |
12 |
was 1GB, and that required NFS mounting, and waiting about 5 days for |
13 |
gcc to build over nfs (was a sparcstation 20). |
14 |
|
15 |
|
16 |
> by this you mean we shouldn't bother trying to create stage1? |
17 |
|
18 |
Stage1's need a seed stage, essentially a stage2 at minimum. I had to |
19 |
use a live gentoo filesystem on my Cobalt RaQ2 to kickstart stage |
20 |
building for the systems (they're mips little endian, at the time, only |
21 |
big endian was supported). Once you can achieve a stage2, then you can |
22 |
work on both stage3 and stage1. I basically tar'ed up my existing |
23 |
stage2 on the cobalt at the time (installed over top a debian |
24 |
installation), and used that as a seed stage for stage1 with catalyst |
25 |
(was a ~100MB+ seed stage vs. ~40MB for stage2's), built a stage1 off |
26 |
that, and then repeated twice more just to make sure I had a pure stage |
27 |
with no debian in it (Took a week). |
28 |
|
29 |
|
30 |
> a lame question here: naturally there would be more poeple working on |
31 |
> this, thus this separate portage tree has to be accessed / written by |
32 |
> several people. how would we achieve this? put it in a CVS somewhere, |
33 |
> and use cvs commit / update? or somehow utilize emerge sync? |
34 |
|
35 |
For this project, and since you already have people working on it, I'd |
36 |
recommend CVS, or some kind of source code control system. It'll make |
37 |
things easier. |
38 |
|
39 |
|
40 |
> I see, catalyst is the name of the game here :) |
41 |
|
42 |
Catalyst is the key for easier gentoo stage building. I always |
43 |
considered the old 'stager' utility nothing more than a glorified bash |
44 |
script, but catalyst is a real tool, and is rather effective at what it |
45 |
does. |
46 |
|
47 |
|
48 |
> when we get there, we'll have to have special precautions for embedded |
49 |
> systems. for example for the Zaurus, you have to flash the new operating |
50 |
> system in, including the kernel and the base filesystem, via a manual |
51 |
> procedure. |
52 |
> |
53 |
> but hey, it's still some way when we have to address this issue :) |
54 |
|
55 |
It's never an easy trip, porting to a new arch is typically only done by |
56 |
one person, then others get interested. tuxus started the mips port |
57 |
that way, tgall_foo is working on ppc64, GMSoft more or less ran the |
58 |
hppa port all by himself for a time before some other people got hppa |
59 |
boxes, iggy is working on an s390 port (unsure how many people are using |
60 |
it), and the original arm port was started by zwelch by himself (he was |
61 |
the one who forked Gentoo into Zynot btw). So in reality, if you've |
62 |
already got a small team together, you're better off from the start than |
63 |
some of the other ports. |
64 |
|
65 |
These precautions you speak of will need to be documented too. One of |
66 |
the people working on the Zaurus stuff may want to maintain a |
67 |
guidexml-based installation guide. Should arm get re-incorporated into |
68 |
gentoo sometime down the road, this will be a required piece, especially |
69 |
to get it integrated into the Gentoo Handbook. |
70 |
|
71 |
|
72 |
--Kumba |
73 |
|
74 |
-- |
75 |
"Such is oft the course of deeds that move the wheels of the world: |
76 |
small hands do them because they must, while the eyes of the great are |
77 |
elsewhere." --Elrond |
78 |
|
79 |
-- |
80 |
gentoo-embedded@g.o mailing list |