1 |
Maciej Mrozowski posted on Thu, 25 Nov 2010 18:38:59 +0100 as excerpted: |
2 |
|
3 |
> On Thursday 25 of November 2010 16:07:45 Alexis Ballier wrote: |
4 |
>> On Thursday 25 November 2010 11:22:39 Paweł Hajdan, Jr. wrote: |
5 |
>> > |
6 |
>> > Could you explain more how the live ebuilds make |
7 |
>> > version bumps easier and less error prone? |
8 |
>> |
9 |
>> By following closely upstream development you are able to adapt the |
10 |
>> live ebuild to the current status, eg, for new deps. [D]oing all these |
11 |
>> changes at once is more error prone than doing it once you notice the |
12 |
>> change for each of them in the live ebuild. |
13 |
>> |
14 |
>> As for version bumps, the live ebuild is simply copied to an ebuild |
15 |
>> with a name matching the version that has just been released. |
16 |
> |
17 |
> It may be worth mentioning it's exactly how all kde-base ebuilds are |
18 |
> prepared. We just svn up in distfiles occasionally, look for changes in |
19 |
> buildsystem and adjust live (stable branch or trunk) ebuilds accordingly |
20 |
> (that being said 4.x.9999 live ebuilds usually reflect the best KDE |
21 |
> version available). |
22 |
> Many Gentoo KDE devs use 4.x.9999 on daily basis (and actually they |
23 |
> should) and thus are able to evaluate patches and take them directly |
24 |
> from disfiles via svn diff -r prev:patchrev to tree if needed. |
25 |
> Ebuilds you see in tree are just generated versions of live ones (blind |
26 |
> copy with replaced keywords, we use bump tool for it however). |
27 |
|
28 |
I had just thought about posting the same thing, from a (kde and x11 |
29 |
overlays user but not normally live ebuild) user perspective. |
30 |
|
31 |
Both those overlays are git-based, and in both cases, I have a script that |
32 |
does a "git --whatchanged --reverse ORIG_HEAD..", that I run after every |
33 |
update. Thus, I follow the overlay git commits quite closely, and see the |
34 |
above in action, even while not actually using the live versions myself. |
35 |
(I did use live for radeon drivers and etc for a bit, when my new card |
36 |
didn't have a released version, even beta or snapshot, with functional |
37 |
OpenGL support, but until 4.5, even so-called stable kde4 was hardly what |
38 |
one would call release status upstream, and I've hesitated to do live |
39 |
there simply because what upstream called stable was better qualified as |
40 |
alpha/beta, broken enough as it was.) |
41 |
|
42 |
While both of these are overlays not in-tree live ebuilds, I can hardly |
43 |
imagine the trouble kde would have trying to do only released version |
44 |
bumps, given the size of the project and the speed at which upstream is |
45 |
changing. xorg isn't quite as bad, but it's transparently obvious it |
46 |
helps there, as well. |
47 |
|
48 |
Also, the experience certainly does have me eager for the main tree git |
49 |
switch-over, as the changelogs are richer and far more useful when I do |
50 |
run into problems, or better yet, avoid them, because I saw the change in |
51 |
the git --whatchanged log (the same biggest reason I read this list, BTW, |
52 |
a heads-up on changes coming down the pike, it's what a responsible |
53 |
sysadmin /does/). |
54 |
|
55 |
Basically, the live ebuild is for package maintainers what the rolling |
56 |
update system is for sysadmin/users. It switches the process from what is |
57 |
effectively a periodic rather opaque big code-dump update, to "rolling |
58 |
updates", allowing the admin/maintainer to manage only a handful of |
59 |
updates at a time, working thru the problems as they appear, when they can |
60 |
still be traced back to individual changes when the inevitable issues do |
61 |
occur, and reacted to accordingly. If the whole rolling update concept |
62 |
makes sense, and it does or Gentoo wouldn't be using it, at least on |
63 |
complex packages/suites with enough change going on to warrant close |
64 |
maintainer attention, live ebuilds closely following upstream's many |
65 |
between-release changes make sense as well. |
66 |
|
67 |
And getting those live ebuilds out there for brave users willing to break |
68 |
their systems to beat on and report bugs on as necessary, certainly |
69 |
doesn't hurt! =:^) |
70 |
|
71 |
-- |
72 |
Duncan - List replies preferred. No HTML msgs. |
73 |
"Every nonfree program has a lord, a master -- |
74 |
and if you use the program, he is your master." Richard Stallman |