1 |
On Fri, 2006-01-06 at 09:00 +0000, Chris Bainbridge wrote: |
2 |
> On 06/01/06, Brian Harring <ferringb@g.o> wrote: |
3 |
> |
4 |
> > Probably better to iron out what y'all actually need and what the dev |
5 |
> > community is willing to put up with. |
6 |
> > |
7 |
> > Eg, would do some research into it, read the archives from last few |
8 |
> > wars over it, in general find (and address) the issues that lead to |
9 |
> > glep19 going still born. |
10 |
> |
11 |
> The problems being: |
12 |
> |
13 |
> 1) Manpower. There are already 10,000 open bugs in bugzilla (and |
14 |
> growing) without adding more. |
15 |
|
16 |
This is probably the primary reason it died. This, of course, ties in |
17 |
greatly to #2. |
18 |
|
19 |
> 2) Lack of interest. Most developers aren't interested in supporting |
20 |
> "old" packages. |
21 |
|
22 |
Again, another of the issues. Another problem was the insistence on |
23 |
using some form of subset of the tree. Using a subset of the tree |
24 |
actually *increases* the workload rather than reduces it. The only real |
25 |
"subset" that can easily work without dramatically increasing workload |
26 |
is to limit to only arch rather than both arch and ~arch. This is |
27 |
simply because it can be scripted. |
28 |
|
29 |
> 3) The enterprise. Both of the above problems would be fixed if |
30 |
> enterprises were contributing developers and/or money. However, they |
31 |
> aren't, so why is that? The truth is most enterprises want to go to a |
32 |
> big company to buy their software. They want one homogeneous binary |
33 |
> system, not a flexible way of building packages from source, and they |
34 |
> want someone else to do it and be responsible for it. |
35 |
|
36 |
While I don't think enterprise support is necessary to accomplish a |
37 |
stable portage tree, it certainly wouldn't hurt. |
38 |
|
39 |
Truthfully, for any "large" enterprise, the company will be maintaining |
40 |
a fair number of their own packages, with custom patches and whatnot. |
41 |
Where I work, we use Red Hat Enterprise Linux. Why? Because we can pay |
42 |
for support. That isn't the point I am going to make here, however. We |
43 |
also have to maintain several hundred RPM packages that either are not |
44 |
included in RHEL or modified by us in some way. What this means is we |
45 |
are now in the business of maintaining a package set, using arguably |
46 |
inferior tools versus ebuilds and portage. The binary support in |
47 |
portage does make it very possible to "build once, deploy everywhere" |
48 |
quite easily. |
49 |
|
50 |
> Look at the arch/~arch split - as a guideline ebuilds are supposed to |
51 |
> move from ~arch -> arch after some reasonable period of time with no |
52 |
> bugs reported (eg. 30 days). Yet as the friendly script shows us, |
53 |
> there are many ebuilds that that stuck in ~arch for a long long time. |
54 |
> Why? The main reason is developer interest - a lot of people only run |
55 |
> ~arch, so the motivation is reduced. It's difficult for users to help |
56 |
> - in the end it's easier to mark stuff ~arch in |
57 |
> /etc/portage/package.keywords than to file a bug requesting that some |
58 |
> developer test and change the ebuild. Proper testing of a tree |
59 |
> requires developers to run both arch and ~arch. The stable proposals |
60 |
> would add yet another tree to test. |
61 |
|
62 |
The current stable proposals do, yes. |
63 |
|
64 |
I'll post up my proposal (in another email), which I believe I had given |
65 |
a couple years ago, somewhere around when GLEP19 was introduced. Given |
66 |
it has been a while, I can't say for certain exactly what order anything |
67 |
came about in, so I won't even try. |
68 |
|
69 |
> The only way I can see to solve these problems is more automation. |
70 |
> Link bugs directly to individual versions (or sets) of ebuilds. Have a |
71 |
> defined process for marking stuff stable - either x days with no bugs, |
72 |
> and/or through sampling of users installed packages to see what is |
73 |
> actually installed out there. Bug voting would fix the disconnect |
74 |
> between users and devs (sometimes people are interested in working on |
75 |
> a random problem to help gentoo - at the moment it's difficult to see |
76 |
> which bugs are really the important ones to users). Decentralise |
77 |
> development - allow users to share their own patches in a searchable |
78 |
> system, rather than force everything through bugzilla; add support to |
79 |
> portage to utilise other peoples trees - the current system of devs |
80 |
> publishing overlays on their web pages/svn and users having to |
81 |
> manually wget or svn up is too difficult for users and removes ebuilds |
82 |
> from the tree - so they are less visible and get less testing. For QA |
83 |
> gentoo really needs a compile farm with automated compile, install and |
84 |
> test (from those ebuilds that support it). Make the system smarter, |
85 |
> instead of throwing more people at the problem. |
86 |
|
87 |
All of these solutions I think are good period, not just to be for some |
88 |
"enterprise" usage, especially multiple repositories (it's coming, yay!) |
89 |
and automated build systems. |
90 |
|
91 |
-- |
92 |
Chris Gianelloni |
93 |
Release Engineering - Strategic Lead |
94 |
x86 Architecture Team |
95 |
Games - Developer |
96 |
Gentoo Linux |