Gentoo Archives: gentoo-dev

From: "Michał Górny" <mgorny@g.o>
To: gentoo-dev@l.g.o, "Anthony G. Basile" <blueness@g.o>
Subject: Re: [gentoo-dev] Inconsistent and messy layout of team maintainership in Gentoo
Date: Thu, 17 Sep 2015 12:54:53
Message-Id: 83643BD2-8D5A-4860-B0BE-70FBCD5EA13E@gentoo.org
In Reply to: Re: [gentoo-dev] Inconsistent and messy layout of team maintainership in Gentoo by "Anthony G. Basile"
1 Dnia 17 września 2015 12:57:08 CEST, "Anthony G. Basile" <blueness@g.o> napisał(a):
2 >On 9/17/15 6:32 AM, Michał Górny wrote:
3 >>
4 >> Dnia 17 września 2015 11:07:43 CEST, "Anthony G. Basile"
5 ><blueness@g.o> napisał(a):
6 >>> On 9/17/15 4:35 AM, Justin (jlec) wrote:
7 >>>> On 16/09/15 23:43, Matthew Thode wrote:
8 >>>>> On 09/16/2015 04:25 PM, Michał Górny wrote:
9 >>>>>> So, what are your thoughts for unmessing this?
10 >>>>>>
11 >>>>> Herds are groups of developers that can then be mapped to a
12 >package.
13 >>>>>
14 >>>> Herds are a group of packages, which are maintained my one or more
15 >>> developers or
16 >>>> a project team consisting of one or more developers.
17 >>>>
18 >>>>
19 >>>>
20 >>> Let's begin by eliminating herds. We can do that by absorbing herds
21 >>> into projects where that makes sense (eg base-system), or expanding
22 >the
23 >>>
24 >>> herd into the respective maintainers where a clear association
25 >between
26 >>> herd and project does not make sense. An ebuild can then have a
27 >>> <project> tag in which case it is maintained by all members of a
28 >>> project
29 >>> or a maintainer list, or a combination of both. It can also have an
30 >>> observer tag for devs that want to track that pkg's maintenance
31 >while
32 >>> not being a maintainer.
33 >> This is not semantically correct for XML. Type of maintainer should
34 >be indicated by subelement or attribute, not three identical elements.
35 >
36 >Are you saying you can't come up with semantic for XML to do this?
37 >C'mon
38 >what I have below isn't the best, but we can come up with something
39 >here.
40
41 I can. Just pointing out the wrongs before some overenthusiastic developer commits it.
42
43 We should really be using the current maintainer element, possibly amended with new attributes. For compatibility with existing tools at least.
44
45 >
46 >>
47 >>> Aliases can be associated with projects in the same way that ebuilds
48 >>> are. An alias can either belong to a project, in which case it
49 >expands
50 >>>
51 >>> to all the project members plus other added emails, or just a free
52 >>> standing alias with just members.
53 >>>
54 >>> Eg1. I'm not on base system, so an pkg with
55 >>>
56 >>> <project>
57 >>> <name>base-system</name>
58 >>> <email>base-system@g.o</email>
59 >> People aren't going to be happy about this. Their VAX-11s may not be
60 >able to handle typing the extra bytes when they do their daily routine
61 >of writing 200 metadata.xml files.
62 >
63 >I'm not sure what you mean here, but the initial migration would be
64 >automated. I don't see why maintenance after that would be difficult.
65
66 Just pointing out the previous outcomes. Gentoo developers aren't happy to sacrifice any kind of laziness for correctness.
67
68 >
69 >>
70 >>> </project>
71 >>> <maintainer>
72 >>> <name>Anthony G. Basile</name>
73 >>> <email>blueness@g.o</email>
74 >>> </maintainer>
75 >>> <observer>
76 >>> <name>Devan Franchini</name>
77 >>> <email>twitch153@gentoo</email>
78 >>> <observer>
79 >> This looks like a serious overkill. Both in the terms of commits and
80 >effort required.
81 >
82 >What's the overkill part? The <observer> tag was just to coalesce the
83 >construction of an alias with the metadata.xml.
84
85 The overkill is that you spam the repo every time someone changes his mind about following a package. Not to mention he needs a developer to commit the change for him.
86
87 >
88 >>
89 >>> would be maintained by all of base system with alias
90 >>> base-system@g.o and me. twitch153 is listed as just an
91 >observer
92 >>>
93 >>> of the pkg and would receive emails but not technically be allowed
94 >to
95 >>> maintain the package. The email alias for this pkg is automatically
96 >>> generated from the metadata.xml.
97 >> This sounds like a lot of alias changes. What about package removals?
98 > Do aliases get removed too?
99 >
100 >Totally rethink the idea of emails aliases as something that is created
101 >
102 >on the fly. We just need to know who should get emails for a package
103 >when it comes to bug reports. Why can't that be calculated on the fly
104 >from the metadata.xml?
105 >
106 >> What about synchronization delays? What about all the extra bugzie
107 >accounts?
108 >Again, I think we can come up with something here. Let's design
109 >something that'll be easy to work with and then figure out how to
110 >implement it.
111
112 Debating pipe dreams without basic understanding of technical aspects is a waste of time.
113
114 Even if we skip the part of implementing mail server changes, and making bugzilla happy about dynamically changing account list, there are major issues left.
115
116 There will be at least the delay between commit in git repo and propagating it to mail server and bugzie. In the meantime, you wouldn't be able to assign and send mail correctly. If it all falls apart, the inability will prolong.
117
118 Then, when package is removed, all assignments immediately become invalid. Mail threads can no longer be continued.
119
120 >
121 >>
122 >>>
123 >>> Eg2. A "free standing" ebuild could have
124 >>>
125 >>> <maintainer>
126 >>> <name>Anthony G. Basile</name>
127 >>> <email>blueness@g.o</email>
128 >>> <proxy/>
129 >>> </maintainer>
130 >>> <maintainer>
131 >>> <email>bugs@××××××××××.nu</email>
132 >>> <name>Johan Bergström</name>
133 >>> </maintainer>
134 >>>
135 >>> Again the alias for this pkg is automatically generated from the
136 >>> metadata.xml
137 >>>
138 >>> Thoughs?
139
140 --
141 Best regards,
142 Michał Górny

Replies