1 |
Its looking like this is the most practical way of going about achieving |
2 |
the stricter version control I and others are thinking about. And its |
3 |
not too much work... but its alot of wasted effort when you have a bunch |
4 |
of people doing it for themselves individually. |
5 |
|
6 |
Maybe it would be a good idea to have a team/project/whatever who's |
7 |
responsibility is to create and maintain snapshots of the Portage tree |
8 |
at different times, and let them take care of assigning Gentoo version |
9 |
numbers and to those snapshots? Meanwhile the rest of the Gentoo team |
10 |
can just keep moving forward with the ever changing metadistribution and |
11 |
not have to worry too much about distilling it for releases so much as |
12 |
doing good work and making cool stuff. |
13 |
|
14 |
Maybe such a project could be a sub-project of stable.gentoo.org, since |
15 |
they seem to be collecting alot of information about stability of things |
16 |
in Portage as it is. |
17 |
|
18 |
I think what you'd end up with would be most of the people working on |
19 |
advancing Gentoo now would be working on what is analagous to -CURRENT |
20 |
in FreeBSD, and then this other group of people would be like the |
21 |
Release Engineering team (looking at drobbins proposal, he already |
22 |
mentioned one), deciding when -CURRENT was ripe for splitting off and |
23 |
stabilizing into a release with such and such goals and featureset. |
24 |
|
25 |
At the least regimented level, it'd be a centralized place for people |
26 |
like Stuart and I to go and pick a static Portage tree to track for our |
27 |
servers. Let the people at that centralized place merge the security |
28 |
updates and bugfixes into the static trees as they see fit. And it |
29 |
shouldn't prove much hindrance to the rest of people working on |
30 |
advancing the bleeding edge, besides maybe modifications to Portage so |
31 |
it has the tree selection abilities built in. |
32 |
|
33 |
On Wed, 25 Jun 2003 21:01:00 +0200 |
34 |
Paul de Vrieze <pauldv@g.o> wrote: |
35 |
|
36 |
> In such a case you might want to run your own cvs ( or subversion) |
37 |
> tree of "sanctioned ebuilds", and instead of emerge sync run cvs |
38 |
> update on the slaves. You then can copy only interesting ebuilds to |
39 |
> the cvs tree, and only wanted changes. Of course this is more work, |
40 |
> but if it should not be too hard to create a "custom tree" based on |
41 |
> the ebuilds that are currently installed. You could put that tree, |
42 |
> with the required distfiles on a custom gentoo bootcd, which you could |
43 |
> use to install all clients. |
44 |
> |
45 |
> Paul |
46 |
|
47 |
-- |
48 |
gentoo-dev@g.o mailing list |