Gentoo Archives: gentoo-dev

From: George Shapovalov <george@g.o>
To: gentoo-dev@g.o
Subject: Re: [gentoo-dev] Ebuilds not getting in :(
Date: Thu, 24 Apr 2003 03:50:55
Message-Id: 200304232050.28318.george@gentoo.org
In Reply to: Re: [gentoo-dev] Ebuilds not getting in :( by Stroller
1 On Wednesday 23 April 2003 17:32, Stroller wrote:
2 > On Tuesday, April 22, 2003, at 02:17 pm, Peter Ruskin wrote:
3 > > On Tuesday 22 Apr 2003 13:59, Frantz Dhin wrote:
4 > >> ... maybe we could make a
5 > >> new keyword? x86 for stable, ~x86 for unstable, and ^x86 for lunatic?
6 > > I couldn't agree more!
7 > Me, either! I don't know that "lunatic" is the best word, but it seems
8 > to me that an additional hierarchy [0] allows for a framework more
9 > flexible & extensible for end lusers. As I understand it builds with the
10 Unfortunately this is not that easy. Just accepting ebuild in and letting them
11 rot is either a dead-end or a security breach (or both :). Think about what
12 to do with them as they get tested and about possible submisisons overlappig
13 already existing *core* ebuilds, yek.. ).
14
15 Please take a look at #1523 to see what's on the plate ;). Only bear in mind,
16 that almost everything in that proposal was written before even KEYWORDS came
17 around, so terminology and, well, pretty much all implementation details are
18 out of date by now.. However the general structure still applies and contains
19 few more (relatively minor as compared to KEYWORDS and gentoo-stats/stable
20 (AKA voting system in that text)) additions.
21
22 I am afraid it is still too early to talk about implementation details (except
23 may be starategic things), as we need to complete the internal restructuring
24 we are attempting right now (and convincing more devs, that we need this kind
25 of thing implemented, as this was not universally accepted yet :)). But the
26 logical structure can and IMHO should be discussed.
27
28 One thing I can already tell for sure, is that security of any such system
29 will be an issue of paramount importance if this kind of thing to be
30 accepted. Namely guarantying by implementation that some unassisted
31 submission does not wreak a havoc on user system no mater what profile that
32 user runs (possibly except "definitely-unstable-you've-been-warned" or
33 whatever it's going to be called :)).
34 There are of course more issues of lesser but still major importance to be
35 considered, such as efficiency on all levels...
36
37 > [1] Am I correctly appreviating "^86, ^PPC or whatever" here? I'm not
38 > doing too well tonight.
39 That'd work, especially if you spell appreviate as abbreviate and make ppc
40 lowercase ;).
41
42 George
43
44 --
45 gentoo-dev@g.o mailing list