1 |
On Sat, Feb 17, 2018 at 10:32 AM, Rich Freeman <rich0@g.o> wrote: |
2 |
> |
3 |
> I've also wondered if a more overlay-driven process might be |
4 |
> beneficial, and certainly others like hasufel have been proponents of |
5 |
> this mode. I can think of some pros and cons: |
6 |
> |
7 |
|
8 |
I will address some of these concerns as best I can. |
9 |
|
10 |
First, if chromium isn't shipping a stable upstream release, then the |
11 |
chromium or "browser-kit" kit can select a version that has undergone some |
12 |
testing and is known to be reliable, and make sure this ends up in the |
13 |
"prime" kit. The latest dev releases can be added to the "dev" kit of |
14 |
'browser-kit'. |
15 |
|
16 |
In Funtoo, right now (since the 1.2 kits are not quite released,) we only |
17 |
have a core-kit that is 1.0-prime. There is no other choice. But for xorg, |
18 |
we have 1.17-prime and then developed 1.19-prime. We continue to backport |
19 |
security fixes into 1.17-prime, but some time ago I flipped the switch to |
20 |
make 1.19-prime the default kit. Funtoo users who did not hard-code the |
21 |
xorg-kit to be 1.17-prime got updated to 1.19, as our ego sync tool |
22 |
switched their default kit over to 1.19-prime. Likewise, we have python-kit |
23 |
3.4-prime and 3.6-prime. And fully support users switching between the two |
24 |
in a seamless fashion. The kits are kept in lock-step as they are generated |
25 |
by a single script, at a single time and our "ego sync" tool ensures that |
26 |
all kits are "aligned" -- this means that you can't use a python-kit |
27 |
3.6-prime generated yesterday with a core-kit generated 30 days ago. This |
28 |
ensures that a catpkg will only exist in one kit at a time, if we happen to |
29 |
move a catpkg, and avoids chaos if the kits are not aligned. (This is easy |
30 |
to implement -- I simply record all the SHA1's for the kits that are |
31 |
generated together, and store these SHA1's in meta-repo -- our 'master kit' |
32 |
that drives the whole process. You can switch kits, but you are locked to |
33 |
the SHA1 that you can choose for each kit.) |
34 |
|
35 |
You can see the metadata I have to drive the kits here: |
36 |
|
37 |
https://github.com/funtoo/meta-repo/blob/master/metadata/kit-info.json |
38 |
https://github.com/funtoo/meta-repo/blob/master/metadata/kit-sha1.json |
39 |
|
40 |
I think this could be quite an exciting collaboration between Funtoo and |
41 |
Gentoo. I know that there has been some interest in getting the |
42 |
multi-profile system into Gentoo as well. |
43 |
|
44 |
And yes, this effort started as finding the best way to take the Gentoo |
45 |
tree and adapting it to our needs and ideal workflow. The upstream work in |
46 |
Gentoo, in the past, has been both a blessing and a curse, because we would |
47 |
suck in changes that would break things in our tree, thus making |
48 |
angry_vincent, well, angry. But our new system provides the buffer we need. |
49 |
No reason why Gentoo couldn't also use this system to provide a buffer. |
50 |
|
51 |
Best, |
52 |
|
53 |
Daniel |