Gentoo Archives: gentoo-dev

From: John Nilsson <john@×××××××.nu>
To: Jon Portnoy <avenj@g.o>
Cc: Jan Schubert <Jan.Schubert@×××.li>, gentoo-dev@l.g.o
Subject: Re: [gentoo-dev] CVS and non-devs (again!)
Date: Thu, 22 Jan 2004 17:34:03
Message-Id: 1074792844.12093.14.camel@newkid.milsson.nu
In Reply to: Re: [gentoo-dev] CVS and non-devs (again!) by Jon Portnoy
1 I must confess my original motivation for this idea. The feature I want
2 is to get ebuild for epia patched packages (xfree, kernel, mythtv and
3 friends) as easy as emerge -u world (mabey even in the same command).
4 Second, puting these ebuild in a semi official tree would expose them to
5 more people, thus hopefully more devs(not gentoo-devs) and testers.
6
7 Higher quality unofficial ebuild in a way...
8
9 I do belive that the issue tracking system is flawed if these kind of
10 bug-reports is a problem though...
11 If all bugs was attached to a specific ebuild, this wouldn't be problem
12 would it?
13
14 -John
15
16
17 On Thu, 2004-01-22 at 17:43, Jon Portnoy wrote:
18 > On Thu, Jan 22, 2004 at 02:21:52PM +0100, Jan Schubert wrote:
19 > > Jon Portnoy wrote:
20 > >
21 > > >And millions of bug reports from users who don't realize they shouldn't
22 > > >report bugs to us on unofficial, unsupported ebuilds, plus users who
23 > > >don't realize Gentoo isn't responsible for any breakage, viruses, or
24 > > >whatever else propogated by an unofficial tree.
25 > > >
26 > > >
27 > > Jon, what John meant is maybe something like to just allow "qualified"
28 > > non devs to support work in the unofficial tree. I'm quite sure that
29 > > these non devs feel very responsible for their work. There might be some
30 > > BugReports in the beginning, but they will be handled by the responsible
31 > > non dev. Maybe some of these non devs will become a dev in the future or
32 > > where asked to become a dev in the past but just don't have the time for
33 > > such a responsible job (personally this is half of the true for me - the
34 > > other half is that i'm feeling that i'm still in the progress of learning).
35 >
36 > If they're not part of the Gentoo organization, there is no
37 > accountability and Gentoo is left holding the bag if someone commits
38 > something that breaks a bunch of people's systems. No, that's not how
39 > things should be, but that's the way things are. As it is we get bug
40 > reports, especially on GNOME stuff, from people using breakmygentoo
41 > ebuilds despite the fact that BMG tells people in about six billion
42 > places not to report bugs to Gentoo.
43 >
44 > >
45 > > On the other side i believe, that users accessing this unofficial tree
46 > > know what they are doing, so it should'nt reflect the official devs that
47 > > much. In my understanding this "feature" is requested by people which
48 > > are not that happy with the current situation (some of them are these
49 > > non devs we talking about). All of them are aware of the consequences.
50 > > This tree would be completely out of scope for "normal" users (they most
51 > > likely never get in touch with this unofficial tree).
52 >
53 > See above. People using BMG ebuilds "know what they're doing" and still
54 > somehow think bugs should go to Gentoo.
55 >
56 > Picking specific people with a history of contributions to commit to
57 > this secondary tree would be silly -- why would that fix anything, if
58 > the current problem is that people feel like they're doing a lot of good
59 > work and not getting picked up as official developers with commit
60 > access? Wouldn't it be the same issue?
61 >
62 > Be realistic. New users are desperate to be "bleeding edge" far moreso
63 > than experienced users and are also more likely to badmouth Gentoo
64 > because of borkage in this secondary tree and report invalid bugs to us,
65 > increasing the workload of bug-wranglers and anyone bugs get mistakenly
66 > assigned to. It would, frankly, lower the overall quality of the
67 > distribution. Bad ebuilds, same programs mistakenly committed under a
68 > different name (or in the wrong category), etc.
69 >
70 > Could committers be given elevated bugzilla privs to handle their own
71 > bugs? Yes. Do I want bug-wranglers wasting their time on handling tons
72 > of bugs on the many low quality ebuilds that would inevitably result?
73 > No.
74 >
75 > And I'm not saying that every ebuild produced by users is
76 > low-quality, but look through a bunch of new ebuilds from various
77 > submitters on bugzilla sometime and you'll find an incredible number of
78 > incomplete or incorrect ebuilds.
79 >
80 > Of course, anyone and their mother can run their own CVS server. But
81 > there's no way I would want something like this hosted on Gentoo
82 > infrastructure or with Gentoo's name on it.
83 >
84 > >
85 > > Maybe all these mails are just a request for an "official" unofficial
86 > > tree!? The problem of lots of unsubmitted ebuilds in bugzilla has to be
87 > > adressed somehow. You should use the motiviatian and the added value
88 > > which these non devs would like to bring in. Don't offend them!
89 >
90 > I'm sorry if refusing to pander to every request is offensive. If people
91 > can't take rejection from time to time, I don't want them working for
92 > Gentoo either.
93 >
94 > Nearly every submitted ebuild is for a fairly small application used by
95 > a fairly small number of people. I don't think putting every single
96 > piece of software ever created in the tree is an urgent issue. QA and
97 > accountability are far more pressing.

Attachments

File name MIME type
signature.asc application/pgp-signature

Replies

Subject Author
Re: [gentoo-dev] CVS and non-devs (again!) Spider <spider@g.o>