ViewVC logotype

Contents of /mitgcm.org/cvspolicy.html

Parent Directory Parent Directory | Revision Log Revision Log | View Revision Graph Revision Graph

Revision 1.2 - (show annotations) (download) (as text)
Sun Jul 2 14:56:16 2000 UTC (21 years, 10 months ago) by cnh
Branch: MAIN
Changes since 1.1: +24 -2 lines
File MIME type: text/html
Formatting edits

1 <html>
2 <head>
3 </head>
4 <body>
5 GCMPACK CVS policies
6 ====================
8 o Introduction
10 This note describes policies that apply to the GCMPACK CVS repository
12 o Why have a policy?
14 CVS itself is a liberal free-for-all product that can be used in a variety
15 of ways. It is designed to provide a system for storing arbitrary files
16 in a way that allows the change history of the individual files to be
17 tracked. If CVS is used without any other policy the result can be a
18 collection of files each of which has complex, multiply branched set of
19 interelated versions. This sort of CVS repository can be come like a
20 library where books are simply stored in a huge heap. Although nothing is
21 actually lost, the task of finding a coherent collection of material soon
22 becomes impossible.
24 The policies we employ address two areas
25 1. Maintaining an orderly and easily identifiable, coherent set of
26 evolving "products".
27 2. Allowing concurrent, on-going development of products.
29 o Development trees and checkpoint trees
31 A directory within the GCMPACK repository resides under either the
32 development branch or the checkpoint branch. Files within each branch
33 follow different policies.
35 o Development tree policies
37 Development trees are intended to be flexible areas where arbitrary files
38 can be stored with multiple versions, many branches supporting multiple
39 ongoing streams of development. Development trees have no policies in
40 place to control complexity. Development trees might be associated with
41 a particular person, a certain project or a particular special piece of
42 work. These trees are intended to be useful areas for storing current
43 work and for archiving partially finished work so that it doesn't get
44 mislaid and s that some record of the development history can be easily
45 maintained. The only policy that applies to development trees is that
46 this style of tree is not intended to be used for providing a
47 "checkpoint" distribution. Tagged configurations of tools built from this
48 style of tree can be distributed, but because these trees do not have any
49 polcies regarding testing of functionality, platform coverage or
50 documentation these trees are not allowed to form the basis of
51 "checkpoint" distrbutions or formal model releases. Other policies can
52 be defined by individuals users of these trees but there are no further
53 global policies. The GCMPACK repository development/ subdirectory is
54 reserved for holding development trees. Development trees also serve as
55 experimental areas for exploring new code management policies.
57 o Checkpoint tree policies
59 Checkpoint trees are intended to provide structured storage areas for
60 holding code that is intended for open distribution and is to be readily
61 downloaded. There are policies governing the operation of these trees
62 which are designed to ensure that distributed codes are clearly
63 identified and meet certain levels of quality.
65 1. Check-out
67 Just do it! Two mechanisms are available. cvsanon for read only access
68 and regular cvs co .... for read/write access.
70 2. Check-in.
72 The code check in procedure for a "checkpoint" tree is as follows
73 2.1 Check out the latest main branch revision.
74 2.2 Merge your changes into that revision.
75 2.3 Build and validate new code.
76 2.4 Check that there have been no further changes to the
77 repository. Repeat from 2.1 if repository has changed.
78 2.5 Get clearance from other developers to check in your changes.
79 2.6 Check in your changed main branch.
80 2.8 Build and validate the new changes.
81 2.9 Tag code as "checkpointNN". Add records to docs/tag-index.
82 2.10 Build and validate test cases (see testing).
83 2.11 Create and install checkpointNN.tar.gz
85 3. Testing
87 Things in a checkpoint tree require a test case that
88 can be used to validate the component.
90 4. Checkpoint tagging
92 No code should be left in limbo. Checking in code and then
93 leaving it in the repository untagged is bad. When you check
94 in code you are creating a new checkpoint. That means you don't
95 check in some code which you "know" works 100% and then go away
96 for two weeks. When you start checking in code you make sure
97 you have time to do the process end-to-end as described in section
98 2.
100 5. Release tagging
102 Releases are only based on checkpoint tree code. Maintenance fixes
103 to releases are also maintained within the checkpoint tree. Files
104 within a release must have accompanying documentation. The form of this
105 documentation depends on the file type.
107 6. Branches
109 Branches are to be used for bug-fixes and code patches to releases
110 only. All other changes e.g. totally new features, bug-fixes to
111 checkpoints are introduced by moving checkpoint levels forward. The
112 only historical code maintenance that is employed is for fixes and
113 patches to formal releases - not checkpoints.
115 o These policies are causing me a big problem, what can I do?
117 The policies are not enforced by any mechanism other than mutual
118 agreement! If you think the policies are not appropriate then let us know
119 and we can discuss changing them. However, if you simply ignore the
120 policies regarding the checkpoint_release trees then your code may be
121 removed and/or your access revoked.
123 o What about bitkeeper
125 We are looking at bitkeeper (www.bitkeeper.com). It looks cool, but
126 policies are still important. Any experience, suggestions let us know.
127 Watch this space!
129 Questions, comments e-mail: gcmpack.code.czars@mitgcm.org
130 </body>
131 </html>

  ViewVC Help
Powered by ViewVC 1.1.22