1 |
Jeremy Olexa wrote: |
2 |
> On Mon, 09 Nov 2009 08:35:57 +0100, Markus Duft <mduft@g.o> wrote: |
3 |
>> Tomasz Pawelczak wrote: |
4 |
>>> Hi, |
5 |
>>> |
6 |
>>> I created a native Soliaris package containing Gentoo Prefix (after |
7 |
>>> fresh bootstrap). You can grab it from |
8 |
>>> |
9 |
> http://mirror.szafa.eu.org/gentoo-prefix-solaris/GPXStage3-v0.2-08.11.2009-i386-x86.pkg.bz2 |
10 |
>>> .It was build and tested on “Solaris 10 11/06 s10x_u3wos_10 X86". |
11 |
>>> After you have installed the package, run |
12 |
>>> “/opt/gentoo/startprefix" to enter Gentoo Prefix, and |
13 |
>>> “emerge –sync" to sync with portage. |
14 |
>> sounds great - even consistent with what i'm doing on windows - i like |
15 |
> it! |
16 |
>> Cheers, Markus |
17 |
> |
18 |
> First, thanks for providing such a solution. |
19 |
|
20 |
++ |
21 |
|
22 |
> |
23 |
> Now, let me bring up the subject of binpkgs. Largely ignored is this: |
24 |
|
25 |
uhhh :) |
26 |
|
27 |
> http://tinderbox.dev.gentoo.org/default-prefix/ - it is historically |
28 |
> "tough" to relocate our prefix binaries, so (IIRC), the binaries that |
29 |
> grobian put up there have a really long path that gets condensed with |
30 |
> chpthtool (paths can only be shrunk, not grown). This is largely preventing |
31 |
|
32 |
and there are additional problems with chpathtool on some platforms. i |
33 |
investigated the whole thing a while ago, since i thought it would be |
34 |
great to (instead of distributing a whole big /opt/gentoo on windows) |
35 |
distribute my gentoo prefix installations split into binpkgs, where the |
36 |
user can decide which to install (much like cygwin does it). i back then |
37 |
also thought about letting the user decide where to install, thats why i |
38 |
was looking at chpathtool. |
39 |
|
40 |
however, the _only_ case where i had real success in changing EPREFIX, |
41 |
is when the new EPREFIX has the _exact_ same length as the old one. |
42 |
seems there is more work to be done one this (and: NO. it was not |
43 |
windows' fault (it most of the time is, i know), since i tried all this |
44 |
on linux ;p) |
45 |
|
46 |
> me from providing binaries for the 6+ prefix arches that I have simply |
47 |
> because I don't have time to maintain my 'production' prefixes in my /home |
48 |
> NFS *and* a super long pathed prefix. Alternatively, I was debating about |
49 |
> making a standard "/opt/gentoo" prefix installation to provide something. |
50 |
> Again, the same boat regarding maintaining multiple prefixes. Although, |
51 |
> that "/opt/gentoo" prefix is a very similar approach to any such stage3 |
52 |
> solution like this. |
53 |
|
54 |
maybe i can still use the binpkgs approach while still hardcoding |
55 |
/opt/gentoo, yes. that would for sure be a cool thing. but it takes away |
56 |
lots of the freedom you get with prefix :(. this would IMHO be a |
57 |
solution for the "i want to try it out" guys, not the power users. and |
58 |
if it's not for everybody, the question is how hard to maintain is it in |
59 |
parallel? |
60 |
|
61 |
on windows, it would not be such a big problem, since i _have_ to |
62 |
maintain the DVD images anyway for my company. if i manage to get time |
63 |
to re-write parts of my setup.exe, i may manage to make it install |
64 |
binpkgs (additionally to a "stage3"?)... |
65 |
|
66 |
Cheers, Markus |
67 |
|
68 |
> |
69 |
> Any brainstorming thoughts? (Excuse the any ignorance on my part with |
70 |
> regards to binpkgs, I'm still a newb in that area :P ) |
71 |
> -Jeremy |
72 |
> |