1 ## This will eventually do to wiki.debian.org/RegressionTestFramework
3 * '''Created''': <<Date(2010-10-07)>>
4 * '''Contributors''': MichaelHanke
5 * '''Packages affected''':
10 This specification describes conventions and tools that allow Debian to
11 distribute and run regression test batteries developed by upstream or
12 Debian developers in a uniform fashion.
16 Ideally software packaged for Debian comes with an exhaustive test suite that
17 can be used to determine whether this software works as expected on the Debian
18 platform. However, especially for complex software, these test suites are often
19 resource hungry (CPU time, memory, diskspace, network bandwidth) and cannot be
20 ran at package build time by buildds. Consequently, test suites are only
21 utilized by the packager on a particular machine, before uploading a new version
24 However, Debian is an integrated system and packaged software is typically made
25 to rely on functionality provided by other Debian packages (e.g. shared
26 libraries) instead of shipping duplicates with different versions in every
27 package -- for many good reasons. Unfortunately, there is also a downside to
28 this: Debian packages often use 3rd-party tools with different versions than
29 those tested by upstream, and moreover, the actual versions might change
30 frequently between to subsequent uploads of a package. Currently a change in a
31 dependency that introduces an incompatibility cannot be detected reliably
32 (before users have filed a bug report) -- even if upstream provides a testsuite
33 that would have caught the breakage. Although there are archive-wide QA efforts
34 (e.g. constantly rebuilding all packages) these tests can only detect API/ABI
35 breakage or functionality tested during build-time checks -- they are not
36 exhaustive for the aforementioned reasons.
38 This is a proposal to, first of all, package upstream test suites in a way that
39 they can be used to run expensive archive-wide QA tests. However, this is also
40 a proposal to establish means to test interactions of software from multiple
41 Debian packages and test proper, continued, integration into the Debian system.
45 * Moritz is a member of the security team. Whenever he applies a patch to fix
46 a security issue he wants to make sure that the behavior of the software
47 remains unchanged. However, in general he only has access to test cases that
48 are included in the source package (if any). In the absence of proper tests
49 he can only either assume that is would work (bad by design), or rely on the
50 respective package maintainer to run the appropriate tests (introduces
51 delays). A packaged exhaustive regression test suite would allow Moritz to
52 perform comprehensive testing on his own and release the fixed package as
53 soon as the tests pass.
55 * Michael is a Debian package maintainer that takes care of three packages
56 each providing a data format conversion utility. While all three tools have
57 their merits there is also lots of overlap. For example, given a particular
58 data file they should all generate identical output. With a Debian
59 regression test framework, Michael can write and package cross-package test
60 suites to ensure that this promise is fulfilled at any time. Moreover,
61 Michael can also develop/package "pipeline" tests that ensure proper
62 functioning of multi-stage/package processing pipelines. For example,
63 testing a whole processing stream from a raw data format conversion, feeding
64 into an analysis to into a visualization package.
66 * Albert is a scientist using Debian for his research activities. The
67 developers of his favorite software tell him to rather use the GreenPants
68 distribution, because they cannot guarantee that their software works
69 properly on Debian. The reason their giving is that Debian has a different
70 version of a numerical library that hasn't been "tested" by the authors.
71 With packaged regression test suites Albert can run, at any given point,
72 a complete test of his Debian system to ensure that everything is working
73 properly given the exact set of base library installed at this very moment.
74 This includes the test suite of the authors of his favorite software, but
75 also all distribution test suites written by Debian developers (see above).
77 * Finally, Lucas likes to run all kinds of tests on all packages in the Debian
78 archive. However, they are mostly concerned with individual packages. A
79 Debian-wide regression test framework would allow Lucas to execute complex
80 tests (suites for individual packages, interoperability tests, or
81 comparative) in an automated fashion, and file bug reports against the
82 respective packages whenever something fails.
86 This specification is applicable to all Debian packages, and Debian as a whole.
90 A specification should be built with the following considerations:
92 * The person implementing it may not be the person writing it. It should be
93 * clear enough for someone to be able to read it and have a clear path
94 * towards implementing it. If it doesn't, it needs more detail.
96 * That the use cases covered in the specification should be practical
97 * situations, not contrived issues.
99 * Limitations and issues discovered during the creation of a specification
100 * should be clearly pointed out so that they can be dealt with explicitly.
102 * If you don't know enough to be able to competently write a spec, you should
103 * either get help or research the problem further. Avoid spending time making
104 * up a solution: base yourself on your peers' opinions and prior work.
106 Specific issues related to particular sections are described further below.
110 The summary should not attempt to say '''why''' the spec is being defined, just
111 '''what''' is being specified.
115 This should be the description of '''why''' this spec is being defined.
117 === Scope and Use Cases ===
119 While not always required, but in many cases they bring much better clarity to
120 the scope and scale of the specification than could be obtained by talking in
123 === Implementation Plan ===
125 This section is usually broken down into subsections, such as the packages
126 being affected, data and system migration where necessary, user interface
127 requirements and pictures (photographs of drawings on paper work well).
131 To implement a specification, the developer should observe the use cases
132 carefully, and follow the design specified. He should make note of places in
133 which he has strayed from the design section, adding rationale describing why
134 this happened. This is important so that next iterations of this specification
135 (and new specifications that touch upon this subject) can use the specification
138 The implementation is very dependent on the type of feature to be implemented.
139 Refer to the team leader for further suggestions and guidance on this topic.
141 == Outstanding Issues ==
143 The specification process requires experienced people to drive it. More
144 documentation on the process should be produced.
146 The drafting of a specification requires english skills and a very good
147 understanding of the problem. It must also describe things to an extent that
148 someone else could implement. This is a difficult set of conditions to ensure
149 throughout all the specifications added.
151 There is a lot of difficulty in gardening obsolete, unwanted and abandoned
152 specifications in the Wiki.
154 == BoF agenda and discussion ==
156 Possible meetings where this specification will be discussed.