Commit 2ee317d975f0544f884e49b35164e9dfe6dde05e
1 parent
46f264fa
Exists in
master
and in
22 other branches
Noosfero development policy
Showing
1 changed file
with
101 additions
and
0 deletions
Show diff stats
... | ... | @@ -0,0 +1,101 @@ |
1 | +# Noosfero Development Policy | |
2 | + | |
3 | +## Developer Roles | |
4 | + | |
5 | +* *Developers* are everyone that are contributing code to Noosfero. | |
6 | +* *Committers* are the people with direct commit access to the Noosfero source | |
7 | + code. They are responsible for reviewing contributions from other developers | |
8 | + and integrating them in the Noosfero code base. They are the members of the | |
9 | + [Noosfero group on Gitlab](https://gitlab.com/groups/noosfero/members). | |
10 | +* *Release managers* are the people that are managing the release of a new | |
11 | + Noosfero version and/or the maintainance work of an existing Noosfero stable | |
12 | + branch. See MAINTAINANCE.md for details on the maintaince policy. | |
13 | + | |
14 | +## Development process | |
15 | + | |
16 | +* Every new feature or non-trivial bugfix should be reviewed by at least one | |
17 | + committer. This must be the case even if the original author is a committer. | |
18 | + | |
19 | + * In the case the original author is a committer, he/she should feel free to | |
20 | + commit directly if after 1 week nobody has provided any kind of feedback. | |
21 | + | |
22 | +* Committers should feel free to push trivial (or urgent) changes directly. | |
23 | + There are no strict rule on what makes a change trivial or urgent; committers | |
24 | + are expected to exercise good judgement on a case by case basis. | |
25 | + | |
26 | +* In the case of unsolvable conflict between commiters regarding any change to | |
27 | + the code, the current release manager(s) will have the final say in the | |
28 | + matter. | |
29 | + | |
30 | +* Release managers are responsible for stablishing a release schedule, and | |
31 | + about deciding when and what to release. | |
32 | + | |
33 | + * Release managers should announce release schedules to the project mailing | |
34 | + lists in advance. | |
35 | + | |
36 | + * The release schedule may include a period of feature freeze, during which | |
37 | + no new features or any other changes that are not pre-approved by the | |
38 | + release manager must be committed to the repository. Committers must | |
39 | + respect the feature freezes. | |
40 | + | |
41 | +## Maintainance process | |
42 | + | |
43 | +### Not all feature releases will be maintained as a stable release | |
44 | + | |
45 | +We will be choosing specific release series to be maintained as stable | |
46 | +releases. | |
47 | + | |
48 | +This means that a given release is not guaranteed to be maintained as a stable | |
49 | +release, but does *not* mean it won't be. Any committer (or anyone, really) can | |
50 | +decide to maintain a given release as stable and seek help from others to do | |
51 | +so. | |
52 | + | |
53 | +### No merges from stable branches to master | |
54 | + | |
55 | +*All* changes must be submitted against the master branch first, and when | |
56 | +applicable, backported to the desired stable releases. Exceptions to this rules | |
57 | +are bug fixes that only apply to a given stable branch and not to master. | |
58 | + | |
59 | +In the past we had non-trivial changes accepted into stable releases while | |
60 | +master was way ahead (e.g. during the rails3 migration period), that made the | |
61 | +merge back into master very painful. By eliminating the need to do these | |
62 | +merges, we save time for the people responsible for the release, and eliminate | |
63 | +the possibility of human errors or oversights causing changes to be accepted | |
64 | +into stable that will be a problem to merge back into master. | |
65 | + | |
66 | +By getting all fixes in master first, we improve the chances that a future | |
67 | +release will not present regressions against bugs that should already be fixed, | |
68 | +but the fixes got lost in a big, complicated merge (and those won't exist | |
69 | +anymore, at least not from stable branches to master). | |
70 | + | |
71 | +After a fix gets into master, backporting changes into a stable release branch | |
72 | +is the responsibility of whoever is maintaing that branch, and those interested | |
73 | +in it. The stable branch release manager(s) are entitled the final say on any | |
74 | +matters related to that branch. | |
75 | + | |
76 | +## Apendix A: how to become a committer | |
77 | + | |
78 | +Every developer that wants to be a committer should create [an issue on | |
79 | +Gitlab](https://gitlab.com/noosfero/noosfero/issues) requesting to be added as | |
80 | +a committer. This request must include information the requestor's previous | |
81 | +contributions to the project. | |
82 | + | |
83 | +If 2 or more commiters consider second the request, the requestor is accepted | |
84 | +as new commiter and added to the Noosfero group. | |
85 | + | |
86 | +The existing committers are free to choose whatever criteria they want to | |
87 | +second the request, but they must be sure that the new committer is a | |
88 | +responsible developer and knows what she/he is doing. They must be aware that | |
89 | +seconding these requests means seconding the actions of the new committer: if | |
90 | +the new committer screw up, her/his seconds screwed up. | |
91 | + | |
92 | +## Apendix B: how to become a release manager | |
93 | + | |
94 | +A new release manager for the development version of Noosfero (i.e. the one | |
95 | +that includes new features, a.k.a. the master branch) is apointed by the | |
96 | +current release manager, and must be a committer first. | |
97 | + | |
98 | +Release managers for stable branches are self-appointed, i.e. whoever takes the | |
99 | +work takes the role. In case of a conflict (e.g. 2+ different people want to | |
100 | +the work but can't agree on working together), the development release manager | |
101 | +decides. | ... | ... |