1 ## This will eventually do to wiki.debian.org/DebTestFramework
3 * '''Created''': <<Date(2010-10-07)>>
4 * '''Contributors''': MichaelHanke, YaroslavHalchenko
5 * '''Packages affected''':
10 This specification describes DebTest -- a framework with conventions and tools
11 that allow Debian to distribute test batteries developed by upstream or Debian
12 developers. DebTest aims to enable developers and users to perform extensive
13 testing of a deployed Debian system or a particular software of interest in a
18 Ideally software packaged for Debian comes with an exhaustive test suite that
19 can be used to determine whether this particular software works as expected on
20 the Debian platform. However, especially for complex software, these test
21 suites are often resource hungry (CPU time, memory, disk space, network
22 bandwidth) and cannot be ran at package build time by buildds. Consequently,
23 test suites are typically utilized manually and only by the respective packager
24 on a particular machine, before uploading a new version to the archive.
26 However, Debian is an integrated system and packaged software typically relies
27 on functionality provided by other Debian packages (e.g. shared libraries)
28 instead of shipping duplicates with different versions in every package -- for
29 many good reasons. Unfortunately, there is also a downside to this: Debian
30 packages often use versions of 3rd-party tools that are different from those
31 tested by upstream, and moreover, the actual versions of dependencies might
32 change frequently between subsequent uploads of a dependent package. Currently
33 a change in a dependency that introduces an incompatibility cannot be detected
34 reliably even if upstream provides a test suite that would have caught the
35 breakage. Therefore integration testing heavily relies on users to detect
36 incorrect functioning and file bug reports. Although there are archive-wide QA
37 efforts (e.g. constantly rebuilding all packages) these tests can only detect
38 API/ABI breakage or functionality tested during build-time checks -- they are
39 not exhaustive for the aforementioned reasons.
41 This is a proposal to, first of all, package upstream test suites in a way that
42 they can be used to run expensive archive-wide QA tests. However, this is also
43 a proposal to establish means to test interactions between software from
44 multiple Debian packages to provide more thorough continued integration and
45 regression testing for the Debian systems.
49 * Moritz is a member of the security team. Whenever he applies a patch to fix
50 a security issue he wants to make sure that the generic behavior of the software
51 remains unchanged. However, in general he only has access to test cases that
52 are included in the source package (if any). In the absence of proper tests
53 he can only either assume that is would work (bad by design), or rely on the
54 respective package maintainer to run the appropriate tests (introduces
55 delays). A packaged exhaustive regression test suite would allow Moritz to
56 perform comprehensive testing on his own and release the fixed package as
57 soon as the tests pass.
59 * Michael is a Debian package maintainer that takes care of three
60 packages each providing a data format conversion utility. While
61 all three tools have their merits there is also lots of
62 overlap. For example, given a particular data file they should all
63 generate identical output. With a DebTest framework, Michael can
64 write and package cross-package test suites to ensure that this
65 promise is fulfilled at any time. Moreover, Michael can also
66 develop/package "pipeline" tests that ensure proper functioning of
67 multi-stage/package processing pipelines (from raw data format
68 conversion to visualization), where some stages could be
69 (re)processed using alternative tools from different software
70 packages promising to provide the same functionality. By testing
71 a whole processing stream while changing the alternative
72 implementations, breakage of the compatibility compliance could be
75 * Yarik is a Debian maintainer of a package where upstream provides
76 a complete analysis pipeline which was used for an article
77 publication. Such analysis requires relatively large array of
78 data and a range of tools from other packages to acquire
79 publication-ready summary of the results. Therefor such analysis
80 cannot be carried out at package build time. Upstream aims to
81 assure the reproducibility of the published results and encourages
82 Yarik to promise correct functioning of the research product on
83 Debian systems. Within the DebTest framework, Yarik can package
84 upstream analysis pipeline along with the target results to assure
85 reproducibility of the scientific findings.
87 * Albert is a scientist using Debian for his research activities. The
88 developers of his favorite software tell him to rather use the GreenPants
89 distribution, because they cannot guarantee that their software works
90 properly on Debian. They reason that Debian has a different
91 version of a numerical library that hasn't been "tested" by the authors.
92 With packaged regression test suites Albert can install and run, at any given point,
93 a complete test of his Debian system to ensure that everything is working
94 properly given the exact set of base libraries installed at this very moment.
95 This includes the test suite of the authors of his favorite software, but
96 also all distribution test suites provided by Debian developers (see above).
98 * Sylvestre maintains a core computational library in Debian.
99 A new version (or other modification) of this library promises performance
100 advantages. Using DebTest he could not only verify the absence of
101 regressions but also to obtain direct performance comparison
102 against the previous version across a range of applications.
104 * Joerg maintains a repository of backports of Debian packages to be
105 installed in a stable environment. He wants to assure that
106 backporting of the packages has not caused a deviation in their
107 intended functioning. By using existing DebTest tests suites he
108 could verify that backported versions of the packages do not break
109 the stability and function as promised within the stable
112 * Mark wants to create a Debian-derived distribution and needs to
113 modify a number of essential packages in order to achieve the desired
114 improvements. He hopes that these changes do not break other Debian
115 packages, but he is not really sure. A comprehensive test battery for the
116 whole Debian system would offer him a way to verify proper functioning
117 of his modified snapshot of Debian -- without having to manually replicate
118 the testing efforts done by thousands of Debian contributors.
120 * Linus is an upstream developer. He just loves the fact that he can tell any
121 of his Debian-based users to just 'apt-get install' something and send him
122 the output of a debtest command, whenever they claim that his software
123 doesn't work properly. It pleases him to see his carefully developed test
124 suite to be conveniently accessible for users.
126 * Finally, Lucas has access to a powerful computing facility and
127 likes to run all kinds of tests on all packages in the Debian archive.
128 A Debian-wide regression test framework would allow Lucas to execute
129 complex test collections (suites for individual packages,
130 interoperability tests, or comparative) in an automated fashion,
131 and file bug reports against the respective packages whenever a
132 malfunction is detected. Some of Lucas friends are not brave enough to file
133 bugs, but still want to contribute. They simply run (selected) tests
134 on their local machines that in turn report results/logs to a Debian
135 dashboard server, where interested parties can get a weather report of
140 This specification is applicable to all Debian packages, and Debian as a whole.
144 A specification should be built with the following considerations:
146 * The person implementing it may not be the person writing it. Specification should be
147 * clear enough for someone to be able to read it and have a clear path
148 * towards implementing it. If it is not straightforward, it needs more detail.
150 * Use cases covered in the specification should be practical
151 * situations, not contrived issues.
153 * Limitations and issues discovered during the creation of a specification
154 * should be clearly pointed out so that they can be dealt with explicitly.
156 * If you don't know enough to be able to competently write a spec, you should
157 * either get help or research the problem further. Avoid spending time making
158 * up a solution: base yourself on your peers' opinions and prior work.
160 Specific issues related to particular sections are described further below.
163 === Core components ===
165 * Organization of the framework
166 - packages might register ways to run basic tests against installed
172 ==== Packaged tests ====
179 * Debug symbols: ....
180 * do not strip symbols from test binary
182 * Packages that register tests might provide a virtual package
183 'test-<packagename>' to allow easy test discovery and retrival via
187 ==== debtest tools ====
190 * single package tests
191 * all (with -f to force even if resources are not sufficient)
192 * tests of dependent packages (discovered via rdepends,
193 "rrecommends" and "rsuggests")
194 * given specific resources demands, just run
195 the ones matching those
196 * Customization/Output::
198 * job resources requirement adjustments
199 . manual customization
200 . request from dashboard for the system (or alike)
202 . local execution (monitor resources)
203 . submit to cluster/cloud
205 . some structured output
206 . interfaces to dashboards
209 ==== Maintainer helpers ====
212 - assess resources/performance:
215 === Supplementary infrastructure ===
217 ==== Dashboard server ====
219 === Implementation Plan ===
221 This section is usually broken down into subsections, such as the packages
222 being affected, data and system migration where necessary, user interface
223 requirements and pictures (photographs of drawings on paper work well).
227 To implement a specification, the developer should observe the use cases
228 carefully, and follow the design specified. He should make note of places in
229 which he has strayed from the design section, adding rationale describing why
230 this happened. This is important so that next iterations of this specification
231 (and new specifications that touch upon this subject) can use the specification
234 The implementation is very dependent on the type of feature to be implemented.
235 Refer to the team leader for further suggestions and guidance on this topic.
237 * Implementation language:
238 - Python unless someone takes the burden to develop
239 and maintain for upcoming years.
241 == Outstanding Issues ==
243 The specification process requires experienced people to drive it. More
244 documentation on the process should be produced.
246 The drafting of a specification requires english skills and a very good
247 understanding of the problem. It must also describe things to an extent that
248 someone else could implement. This is a difficult set of conditions to ensure
249 throughout all the specifications added.
251 There is a lot of difficulty in gardening obsolete, unwanted and abandoned
252 specifications in the Wiki.
254 == BoF agenda and discussion ==
256 Possible meetings where this specification will be discussed.