1 |
Thanks for the warning Alec. Since I already have very limited |
2 |
knowledge of the internals, I figure this will be a good trial by fire |
3 |
(learn sort of the portage internals and issues with prefixed |
4 |
installs). If it works, perhaps I will be able to devote more time to |
5 |
it and do more than a simple prototype (which might involve writing a |
6 |
spec, talking to Fink people and seeing how they resolved these sorts |
7 |
of issues and all that fun stuff). Right now, I'm not very qualified |
8 |
to write a spec, but I figure when I get my hands into it I'll start |
9 |
seeing a lot more of the issues one is bound to run into. If it's so |
10 |
hard that's it's impossible then maybe I should just give up now.... |
11 |
What sort of "plan" do you suggest? |
12 |
|
13 |
On 10/5/05, Alec Warner <warnera6@×××××××.edu> wrote: |
14 |
> Brian Harring wrote: |
15 |
> |
16 |
> >On Wed, Oct 05, 2005 at 11:31:32PM +0100, Ciaran McCreesh wrote: |
17 |
> > |
18 |
> > |
19 |
> >>On Wed, 5 Oct 2005 16:13:06 -0500 Brian Harring <ferringb@g.o> |
20 |
> >>wrote: |
21 |
> >>| A) would like to hear what you think is required planning wise |
22 |
> >>| compared to the previous haubi prototype patch. |
23 |
> >> |
24 |
> >>There has been no serious discussion on how *ebuilds* will use the |
25 |
> >>prefix system. Hacking econf and expecting PREFIX to be sufficient is |
26 |
> >>naive from a tree-perspective. |
27 |
> >> |
28 |
> >> |
29 |
> > |
30 |
> >econf isn't the only change required; the point is that whatever is |
31 |
> >decided, would have to be added to econf thus covering a good chunk of |
32 |
> >ebuilds in the tree that don't require fancy voodoo. |
33 |
> > |
34 |
> >The basic proposal of haubi's glep (ignoring the portage innard |
35 |
> >modifications) came down to addition of a prefix var, that would be |
36 |
> >required slipped in for any fs installation paths (--prefix=$PREFX |
37 |
> >fex). |
38 |
> > |
39 |
> >Beyond that, there is the shebang issue which can be addresses via a |
40 |
> >combination of automated scans/fixes, and fixing bugs as it's hit. |
41 |
> >Hardcoded vars in scripts for the path to a binary are an issue also, |
42 |
> >although again, scans can be done to at least check for it. |
43 |
> > |
44 |
> >Leaves mangling the build process so that the build framework of the |
45 |
> >package uses the prefix offset files, rather then / . For c/c++ |
46 |
> >source, usual trick from fink afaik involves a mangling of cflags with |
47 |
> >-I tacked in. Kinda ugly, although I'd expect there is a better |
48 |
> >route. |
49 |
> > |
50 |
> >Packages that pull include/compile settings/args from a utility |
51 |
> >(thinking python configuration tools, and pkgconfig) shouldn't be too |
52 |
> >horrid to change, since it's a matter of modifying it in one place |
53 |
> >(theoretically :). |
54 |
> >~harring |
55 |
> > |
56 |
> > |
57 |
> > |
58 |
> > |
59 |
> I guess in the end trying to do something like this is a difficult |
60 |
> process. I am wary of anyone who wants to just jump into an application |
61 |
> like portage and just magically write in this kind of support. In the |
62 |
> end one could just try and go step by step, but nothing guarantee's you |
63 |
> won't miss something, or because it works with packages x,y,z that it |
64 |
> will work for all packages. |
65 |
> |
66 |
> If you have two weeks to do it in, I wish you the best of luck. Maybe |
67 |
> you are good enough at learning portage internals to get it done, but |
68 |
> even after portage support is done there are still plenty of other factors. |
69 |
> |
70 |
> In the end I side with Ciaran on this one. You need to know all the |
71 |
> bases to cover here in order to make this work well. Going ahead with |
72 |
> no plan is stupid IMHO. |
73 |
> -- |
74 |
> gentoo-portage-dev@g.o mailing list |
75 |
> |
76 |
> |
77 |
|
78 |
-- |
79 |
gentoo-portage-dev@g.o mailing list |