1 |
> |
2 |
> Thanks for all the work guys! |
3 |
> |
4 |
> However, for a jumbo patch like this, I find the "documentation" quite |
5 |
> limited. I admit I stopped reading after a page or 3, since it started |
6 |
> to get too scary to think of just applying it. I certainly need more |
7 |
> talking on this, and given the huge size, I'd even opt for seeing if we |
8 |
> can push stuff directly to trunk, such that I don't have to shadow |
9 |
> maintain it. |
10 |
|
11 |
Attached is (for those interested) a small doc I wrote up. The patch is checked-in (conditional with a USE flag...). |
12 |
|
13 |
As for integrating parts of it into main: one _could_ change the patch, so that a more generic ROOTobject class is introduced, which then could be extended in the prefix branch to contain the prefix too. However this would be quite some work on both main-trunk and prefix-trunk... |
14 |
|
15 |
Cheers, Markus |
16 |
|
17 |
> |
18 |
> |
19 |
> On 11-09-2008 11:17:07 +0200, Markus Duft wrote: |
20 |
> > > |
21 |
> > > Hi! |
22 |
> > |
23 |
> > Right after clicking "send" I found a bug in the patch which I |
24 |
> introduced |
25 |
> > while cleaning it up for submission :) attached is now a working |
26 |
> version, |
27 |
> > which has just a small change: the setting of config_root is now done |
28 |
> > differently, since the config class can be instantiated multiple |
29 |
> times for |
30 |
> > different prefix installations... |
31 |
> > |
32 |
> > Cheers, Markus |
33 |
> > |
34 |
> > > |
35 |
> > > Some time ago I talked on this list about my plans on using portage |
36 |
> to |
37 |
> > > emerge packages for native Windows under Interix. To be able to do |
38 |
> > > this, haubi and I had to "help" portage a little :) we now have a |
39 |
> patch |
40 |
> > > ready with which this is possible. Not only windows draws |
41 |
> advantages |
42 |
> > > out of the patch, but now it is possible to create a "child" |
43 |
> prefix. |
44 |
> > > You can create a base EPREFIX, which contains a system that doesn't |
45 |
> > > change too often, and then merge packages to different prefixes |
46 |
> > > building upon that parent prefix. |
47 |
> > > |
48 |
> > > For now, only DEPEND's are resolved from the parent EPREFIX -- |
49 |
> RDEPEND |
50 |
> > > and PDEPEND *must* be installed in the child EPREFIX. If a DEPEND |
51 |
> > > cannot be merged into the parent prefix, portage tries to merge is |
52 |
> to |
53 |
> > > the child. |
54 |
> > > |
55 |
> > > Maybe some of you can have a brief look at the patch, and if there |
56 |
> are |
57 |
> > > no objections, I'd like to add the patch to the portage ebuild |
58 |
> > > conditionally with a USE flag for more testing. If it does not |
59 |
> destroy |
60 |
> > > anything (I greatly doubt that it does...), it would be great to |
61 |
> see it |
62 |
> > > going into svn... |
63 |
> > > |
64 |
> > > P.S.: haubi wrote up some comments on the patch and other things... |
65 |
> I |
66 |
> > > attached them as-is, so ask haubi :) |
67 |
> > > P.S.S.: the patch is against anonsvn-checked-out trunk of prefix |
68 |
> > > branch... |
69 |
> > > |
70 |
> > > Cheers, Markus |
71 |
> |
72 |
> |
73 |
> |
74 |
> |
75 |
> -- |
76 |
> Fabian Groffen |
77 |
> Gentoo on a different level |