1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
|
Samba maintainers
-----------------
This file lists the maintainers for subsystems in Samba. Please see
the end of the file for information on how the maintainers system
works. If you can't work out who the maintainer is for some code,
please ask on the samba-technical list or on the samba-technical IRC
channel.
=======================================================================
directory: lib/tevent/
maintainers:
Stefan Metzmacher <metze@samba.org>
policy:
All commits require review by the maintainer.
If no maintainer is available for longer than a week
discussion on the samba-technical list and review by 2
Samba-Team members is needed (e.g. Andrew Tridgell <tridge@samba.org>
and Volker Lendecke <vl@samba.org>).
Larger changes need also discussion on the samba-technical list
and review by all maintainers.
directory: lib/tsocket/
maintainers:
Stefan Metzmacher <metze@samba.org>
policy:
All commits require review by the maintainer.
If no maintainer is available for longer than a week
discussion on the samba-technical list and review by 2
Samba-Team members is needed.
Larger changes need also discussion on the samba-technical list
and review by all maintainers.
files: buildtools/**, source4/**/wscript
maintainers:
Andrew Tridgell <tridge@samba.org>
Jelmer Vernooij <jelmer@samba.org>
policy:
small commits to master allowed if all existing tests
pass. Larger commits require discussion on the samba-technical
list and review by the maintainer
files: lib/tdb
maintainers:
Rusty Russell <rusty@samba.org>
policy:
Mail/CC changes to the maintainer, commit the changes
unless the maintainer objects.
files: lib/talloc
maintainers:
Andrew Tridgell <tridge@samba.org>
Rusty Russell <rusty@samba.org>
policy:
small commits to master allowed if all existing tests
pass. Larger commits require discussion on samba-technical
list and review by the maintainer
files: lib/tevent/py*, lib/talloc/py*, source4/lib/ldb/py*, lib/tdb/py*
maintainers:
Jelmer Vernooij <jelmer@samba.org>
policy:
Larger commits require pre-push review by the maintainer or
one of the maintainers of the containing subsystem.
Other non-trivial (typo, etc) commits require pre- or post-push review by the
maintainer or one of the maintainers of the containing subsystem.
=======================================================================
Samba Maintainers System
------------------------
The Samba project has adopted a maintainers system, with the following
approach:
- we have created a new 'MAINTAINERS.txt' file in the root of the git
tree
- that file will contain a list of subsystems, and along with each
subsystem a list of maintainers
- subsystems may be subdirectories, or logical groups of files (for
example "build system" or "selftest" could be subsystems that span
multiple directories)
- if a subsystem is not listed in the MAINTAINERS.txt file, then this
maintainers proposal does not apply to that subsystem. The previous
Samba development methods apply to unlisted subsystems.
- when we first create the MAINTAINERS.txt it will be empty, thus on
the first day of adoption there is no actual change to our
development practices
- we will add subsystems to the MAINTAINERS.txt file via consensus
within the Samba Team. This means that someone would propose
themselves, or another team member, as a subsystem maintainer, and
if there are no objections then they can push a change to the
maintainers file after a couple of days waiting for replies. If
there is an existing maintainer for that subsystem then at minimum
the person proposing should wait for a positive ack from the
previous maintainer.
- a typical subsystem declaration would be:
directory: /libds
maintainers:
Andrew Bartlett <abartlet@samba.org>
Andrew Tridgell <tridge@samba.org>
policy:
small commits to master allowed if all existing tests
pass. Larger commits require discussion on samba-technical
list and review by the maintainer
- the maintainers for a subsystem may update the policy for that
subsystem at any time by pushing a commit to the MAINTAINERS.txt
file. Significant changes should also be sent to the
samba-technical list to ensure that all developers are aware of the
policy change
- a subsystem may have multiple maintainers, and it is expected that
this will be the case for many of our subsystems.
- a maintainer may delegate responsibility to someone else for a
period of time (such as during rapid development or when the
maintainer is away). A maintainer may also appoint a backup
maintainer. These changes should be noted in the maintainers file,
and removed when no longer relevent.
- maintainer handover would happen by agreement between the old and
new maintainer, and is signified by a commit to the MAINTAINERS.txt
file. If agreement cannot be reached then we can resolve the
disagreement using discussions on the team list. If agreement still
can't be reached then the maintainer won't change.
What does it mean to be a maintainer?
-------------------------------------
If you are a maintainer for a subsystem then you have some additional
rights and responsibilies for that code. Specifically:
- you should make time to review any proposed changes to any
subsystems that you maintain. You should then provide feedback on
proposed changes or sign off on the changes once you are happy with
them.
- you may choose the policy for the subsystems you maintain. That
policy could be a permissive one, where you allow for small changes
without review, or it could be a strict one, where you only allow
reviewed changes to be pushed.
- being a maintainer for a subsystem does not override the "right of
veto" of other team members for technical objections. See the
"right of veto" section below for more information.
- the maintainers can set the developmental direction of the
subsystem, but should strive to achieve concensus where possible
with other team members for the benefit of the whole
project.
Note that if you set a permissive policy on your subsystem, so that
small changes may be pushed without review, you are still responsible
for reviewing changes if someone specifically asks you to review a
patch.
Try to reuse policy wording
---------------------------
It would be good if we end up with only a few sets of policy wording,
rather than a completely different policy for each subsystem. To try
to achieve that, maintainers should try to re-use an existing policy
wording if possible.
The right of veto
-----------------
Over the last few years the Samba Team has started to use a +1/-1
voting system, which was inspired by the Apache voting system for
technical issues (see http://www.apache.org/foundation/voting.html).
For the maintainers proposal to work, I think we need to ensure that
everyone understands what a -1 "veto" vote means on a technical issue.
For purely technical issues, the +1/-1 voting system should not be a
"most votes wins" system. Instead a single -1 vote is supposed to
override any number of +1 votes, so a -1 vote is a "veto", and all
team members have the right to give a -1 veto vote on any purely
technical issue.
Along with the right to give a -1 veto vote comes the responsibility
to backup that veto with a technical argument, and the willingness to
then defend your argument in any subsequent discussions and to work
with the patch proposer to find a solution. If you do not backup your
-1 veto vote, or you are unwilling on unable to participate in any
discussions that arise from that veto, then the veto vote may be
disregarded.
Note that a veto is supposed to be used only for purely technical
reasons, so for example pointing out a security concern with a change,
or pointing out that the code may segfault or cause a regression of
functionality.
|
