3 In our current testing environment, we spend a significant amount of effort
4 crafting end-to-end tests for error conditions that could easily be captured by
5 unit tests (or we simply forgo some hard-to-setup and rare error conditions).
6 Unit tests additionally provide stability to the codebase and can simplify
7 debugging through isolation. Writing unit tests in pure C, rather than with our
8 current shell/test-tool helper setup, simplifies test setup, simplifies passing
9 data around (no shell-isms required), and reduces testing runtime by not
10 spawning a separate process for every test invocation.
12 We believe that a large body of unit tests, living alongside the existing test
13 suite, will improve code quality for the Git project.
17 For the purposes of this document, we'll use *test framework* to refer to
18 projects that support writing test cases and running tests within the context
19 of a single executable. *Test harness* will refer to projects that manage
20 running multiple executables (each of which may contain multiple test cases) and
21 aggregating their results.
23 In reality, these terms are not strictly defined, and many of the projects
24 discussed below contain features from both categories.
26 For now, we will evaluate projects solely on their framework features. Since we
27 are relying on having TAP output (see below), we can assume that any framework
28 can be made to work with a harness that we can choose later.
33 We believe the best way forward is to implement a custom TAP framework for the
34 Git project. We use a version of the framework originally proposed in
35 https://lore.kernel.org/git/c902a166-98ce-afba-93f2-ea6027557176@gmail.com/[1].
37 See the <<framework-selection,Framework Selection>> section below for the
38 rationale behind this decision.
41 == Choosing a test harness
43 During upstream discussion, it was occasionally noted that `prove` provides many
44 convenient features, such as scheduling slower tests first, or re-running
45 previously failed tests.
47 While we already support the use of `prove` as a test harness for the shell
48 tests, it is not strictly required. The t/Makefile allows running shell tests
49 directly (though with interleaved output if parallelism is enabled). Git
50 developers who wish to use `prove` as a more advanced harness can do so by
51 setting DEFAULT_TEST_TARGET=prove in their config.mak.
53 We will follow a similar approach for unit tests: by default the test
54 executables will be run directly from the t/Makefile, but `prove` can be
55 configured with DEFAULT_UNIT_TEST_TARGET=prove.
58 [[framework-selection]]
59 == Framework selection
61 There are a variety of features we can use to rank the candidate frameworks, and
62 those features have different priorities:
64 * Critical features: we probably won't consider a framework without these
65 ** Can we legally / easily use the project?
66 *** <<license,License>>
67 *** <<vendorable-or-ubiquitous,Vendorable or ubiquitous>>
68 *** <<maintainable-extensible,Maintainable / extensible>>
69 *** <<major-platform-support,Major platform support>>
70 ** Does the project support our bare-minimum needs?
71 *** <<tap-support,TAP support>>
72 *** <<diagnostic-output,Diagnostic output>>
73 *** <<runtime-skippable-tests,Runtime-skippable tests>>
74 * Nice-to-have features:
75 ** <<parallel-execution,Parallel execution>>
76 ** <<mock-support,Mock support>>
77 ** <<signal-error-handling,Signal & error-handling>>
79 ** <<project-kloc,Project KLOC>>
80 ** <<adoption,Adoption>>
85 We must be able to legally use the framework in connection with Git. As Git is
86 licensed only under GPLv2, we must eliminate any LGPLv3, GPLv3, or Apache 2.0
89 [[vendorable-or-ubiquitous]]
90 === Vendorable or ubiquitous
92 We want to avoid forcing Git developers to install new tools just to run unit
93 tests. Any prospective frameworks and harnesses must either be vendorable
94 (meaning, we can copy their source directly into Git's repository), or so
95 ubiquitous that it is reasonable to expect that most developers will have the
96 tools installed already.
98 [[maintainable-extensible]]
99 === Maintainable / extensible
101 It is unlikely that any pre-existing project perfectly fits our needs, so any
102 project we select will need to be actively maintained and open to accepting
103 changes. Alternatively, assuming we are vendoring the source into our repo, it
104 must be simple enough that Git developers can feel comfortable making changes as
105 needed to our version.
107 In the comparison table below, "True" means that the framework seems to have
108 active developers, that it is simple enough that Git developers can make changes
109 to it, and that the project seems open to accepting external contributions (or
110 that it is vendorable). "Partial" means that at least one of the above
113 [[major-platform-support]]
114 === Major platform support
116 At a bare minimum, unit-testing must work on Linux, MacOS, and Windows.
118 In the comparison table below, "True" means that it works on all three major
119 platforms with no issues. "Partial" means that there may be annoyances on one or
120 more platforms, but it is still usable in principle.
125 The https://testanything.org/[Test Anything Protocol] is a text-based interface
126 that allows tests to communicate with a test harness. It is already used by
127 Git's integration test suite. Supporting TAP output is a mandatory feature for
128 any prospective test framework.
130 In the comparison table below, "True" means this is natively supported.
131 "Partial" means TAP output must be generated by post-processing the native
134 Frameworks that do not have at least Partial support will not be evaluated
137 [[diagnostic-output]]
138 === Diagnostic output
140 When a test case fails, the framework must generate enough diagnostic output to
141 help developers find the appropriate test case in source code in order to debug
144 [[runtime-skippable-tests]]
145 === Runtime-skippable tests
147 Test authors may wish to skip certain test cases based on runtime circumstances,
148 so the framework should support this.
150 [[parallel-execution]]
151 === Parallel execution
153 Ideally, we will build up a significant collection of unit test cases, most
154 likely split across multiple executables. It will be necessary to run these
155 tests in parallel to enable fast develop-test-debug cycles.
157 In the comparison table below, "True" means that individual test cases within a
158 single test executable can be run in parallel. We assume that executable-level
159 parallelism can be handled by the test harness.
164 Unit test authors may wish to test code that interacts with objects that may be
165 inconvenient to handle in a test (e.g. interacting with a network service).
166 Mocking allows test authors to provide a fake implementation of these objects
167 for more convenient tests.
169 [[signal-error-handling]]
170 === Signal & error handling
172 The test framework should fail gracefully when test cases are themselves buggy
173 or when they are interrupted by signals during runtime.
178 The size of the project, in thousands of lines of code as measured by
179 https://dwheeler.com/sloccount/[sloccount] (rounded up to the next multiple of
180 1,000). As a tie-breaker, we probably prefer a project with fewer LOC.
185 As a tie-breaker, we prefer a more widely-used project. We use the number of
186 GitHub / GitLab stars to estimate this.
191 :true: [lime-background]#True#
192 :false: [red-background]#False#
193 :partial: [yellow-background]#Partial#
195 :gpl: [lime-background]#GPL v2#
196 :isc: [lime-background]#ISC#
197 :mit: [lime-background]#MIT#
198 :expat: [lime-background]#Expat#
199 :lgpl: [lime-background]#LGPL v2.1#
201 :custom-impl: https://lore.kernel.org/git/c902a166-98ce-afba-93f2-ea6027557176@gmail.com/[Custom Git impl.]
202 :greatest: https://github.com/silentbicycle/greatest[Greatest]
203 :criterion: https://github.com/Snaipe/Criterion[Criterion]
204 :c-tap: https://github.com/rra/c-tap-harness/[C TAP]
205 :check: https://libcheck.github.io/check/[Check]
206 :clar: https://github.com/clar-test/clar[Clar]
208 [format="csv",options="header",width="33%",subs="specialcharacters,attributes,quotes,macros"]
210 Framework,"<<license,License>>","<<vendorable-or-ubiquitous,Vendorable or ubiquitous>>","<<maintainable-extensible,Maintainable / extensible>>","<<major-platform-support,Major platform support>>","<<tap-support,TAP support>>","<<diagnostic-output,Diagnostic output>>","<<runtime--skippable-tests,Runtime- skippable tests>>","<<parallel-execution,Parallel execution>>","<<mock-support,Mock support>>","<<signal-error-handling,Signal & error handling>>","<<project-kloc,Project KLOC>>","<<adoption,Adoption>>"
211 {custom-impl},{gpl},{true},{true},{true},{true},{true},{true},{false},{false},{false},1,0
212 {greatest},{isc},{true},{partial},{true},{partial},{true},{true},{false},{false},{false},3,1400
213 {criterion},{mit},{false},{partial},{true},{true},{true},{true},{true},{false},{true},19,1800
214 {c-tap},{expat},{true},{partial},{partial},{true},{false},{true},{false},{false},{false},4,33
215 {check},{lgpl},{false},{partial},{true},{true},{true},{false},{false},{false},{true},17,973
216 {clar},{isc},{false},{partial},{true},{true},{true},{true},{false},{false},{true},1,192
219 === Additional framework candidates
221 Several suggested frameworks have been eliminated from consideration:
223 * Incompatible licenses:
224 ** https://github.com/zorgnax/libtap[libtap] (LGPL v3)
225 ** https://cmocka.org/[cmocka] (Apache 2.0)
226 * Missing source: https://www.kindahl.net/mytap/doc/index.html[MyTap]
228 ** https://nemequ.github.io/munit/[µnit]
229 ** https://github.com/google/cmockery[cmockery]
230 ** https://github.com/lpabon/cmockery2[cmockery2]
231 ** https://github.com/ThrowTheSwitch/Unity[Unity]
232 ** https://github.com/siu/minunit[minunit]
233 ** https://cunit.sourceforge.net/[CUnit]
238 * Add useful tests of library-like code
240 https://lore.kernel.org/git/20230502211454.1673000-1-calvinwan@google.com/[stdlib
242 * Run alongside regular `make test` target