OLD | NEW |
---|---|
(Empty) | |
1 # Field Trial Testing Configuration | |
2 | |
3 This directory contains the field trial configuration used to ensure testing | |
4 coverage of the experiments defined in `fieldtrial_testing_config.json`. | |
5 | |
6 Note that these apply specifically for Chromium builds. Chrome branded / | |
7 official builds do not use these definitions. | |
8 | |
9 ## Config File Format | |
10 | |
11 ```json | |
12 { | |
13 "StudyName": [ | |
14 { | |
15 "platforms": [Array of Strings of Valid Platforms for These Experime nts], | |
16 "experiments": [ | |
17 { | |
18 "//0:": "Comment Line 0. Lines 0-9 are supported.", | |
19 "name": "ExperimentName", | |
20 "params": {Dictionary of Params}, | |
21 "enable_features": [Array of Strings of Features], | |
22 "disable_features": [Array of Strings of Features] | |
23 }, | |
24 ... | |
25 ] | |
26 }, | |
27 ... | |
28 ], | |
29 ... | |
30 } | |
31 ``` | |
32 | |
33 The config file is a dictionary at the top level mapping a study name to an | |
34 array of *experiment descriptions*. The study name should match the Field Trial | |
35 study name. | |
36 | |
37 ### Experiment Descriptions | |
38 | |
39 Each *experiment description* is a dictionary containing both `platform` and | |
40 `experiments`. | |
41 | |
42 `platform` is an array of strings of valid platforms and the | |
43 strings may be `android`, `chromeos`, `ios`, `linux`, `mac`, or `windows`. | |
44 | |
45 `experiments` is an array containing the *experiments*. | |
46 | |
47 The converter uses the platforms array to determine what experiments to include | |
48 for the study. All matching platforms will have their experiments concatenated | |
49 together for the study. | |
50 | |
51 The first available experiment after platform filtering and concatenation is the | |
52 default experiment for Chromium builds. This experiment is also used for perf | |
53 bots and browser tests in the waterfall. | |
Alexei Svitkine (slow)
2016/09/21 21:38:17
I feel like this paragraph should live in the firs
robliao
2016/09/21 22:06:09
Done.
| |
54 | |
55 ### Experiments | |
56 Each *experiment* is a dictionary that must contain the `name` key, identifying | |
57 the experiment group name which should match the Field Trial experiment group | |
58 name. | |
59 | |
60 The remaining keys, `params`, `enable_features`, and `disable_features` are | |
61 optional. | |
62 | |
63 `params` is a dictionary mapping parameter name to parameter. | |
64 | |
65 `enable_features` and `disable_features` indicate which features should be | |
66 enabled and disabled respectively. As a reminder, as the variations framework | |
67 does not actually fetch the Field Trial definitions from the server for Chromium | |
68 builds, so any feature enabling or disabling must be done here. | |
69 | |
70 #### Comments | |
71 | |
72 Each experiment may have up to 10 lines of comments. The comment key must be of | |
73 the form `//N` where `N` is between 0 and 9. | |
74 | |
75 ```json | |
76 { | |
77 "AStudyWithExperimentComment": [ | |
78 { | |
79 "platforms": ["chromeos", "linux", "mac", "win"], | |
80 "experiments": [ | |
81 { | |
82 //0: "This is the first comment line.", | |
Alexei Svitkine (slow)
2016/09/21 21:38:17
The syntax seems different here than on line 18. P
robliao
2016/09/21 22:06:09
Made it consistent, although this change would be
| |
83 //1: "This is the second comment line.", | |
84 "name": "DesktopExperiment" | |
85 } | |
86 ] | |
87 } | |
88 ] | |
89 } | |
90 ``` | |
91 | |
92 ### Specifying Different Experiments for Different Platforms | |
93 Simply specify two different experiment configurations under the same study: | |
94 | |
95 ```json | |
96 { | |
97 "DifferentExperimentsPerPlatform": [ | |
98 { | |
99 "platforms": ["chromeos", "linux", "mac", "win"], | |
100 "experiments": [{ "name": "DesktopExperiment" }] | |
101 }, | |
102 { | |
103 "platforms": ["android", "ios"], | |
104 "experiments": [{ "name": "MobileExperiment" }] | |
105 } | |
106 ] | |
107 } | |
108 ``` | |
109 | |
110 ## Presubmit | |
111 The presubmit tool will ensure that your changes follow the correct ordering and | |
112 format. | |
OLD | NEW |