Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
Property-based Testing is a Mindset
Search
Andrea Leopardi
December 06, 2017
Programming
0
690
Property-based Testing is a Mindset
Andrea Leopardi
December 06, 2017
Tweet
Share
More Decks by Andrea Leopardi
See All by Andrea Leopardi
gen_statem - OTP's Unsung Hero
whatyouhide
1
190
The World is a Network (and We Are Just Nodes)
whatyouhide
1
200
BEAM: The Perfect Fit for Networks
whatyouhide
1
180
Update from the Elixir team - 2022
whatyouhide
0
390
Testing Asynchronous OTP
whatyouhide
1
510
Elixir Sightseeing Tour
whatyouhide
0
410
Mint - Disrupting HTTP clients
whatyouhide
0
240
BEAM Architecture Handbook
whatyouhide
7
2.8k
The Evolution of a Language
whatyouhide
0
140
Other Decks in Programming
See All in Programming
小さく段階的リリースすることで深夜メンテを回避する
mkmk884
2
150
Signal-Based Data FetchingWith the New httpResource
manfredsteyer
PRO
0
130
Preact、HooksとSignalsの両立 / Preact: Harmonizing Hooks and Signals
ssssota
1
1.2k
PHPer's Guide to Daemon Crafting Taming and Summoning
uzulla
2
1.1k
Going Structural with Named Tuples
bishabosha
0
190
DomainException と Result 型で作る型安全なエラーハンドリング
karszawa
0
850
CRE Meetup!ユーザー信頼性を支えるエンジニアリング実践例の発表資料です
tmnb
0
580
Go1.24で testing.B.Loopが爆誕
kuro_kurorrr
0
180
snacks.nvim内のセットアップ不要なプラグインを紹介 / introduce_snacks_nvim
uhooi
0
370
Rollupのビルド時間高速化によるプレビュー表示速度改善とバンドラとASTを駆使したプロダクト開発の難しさ
plaidtech
PRO
1
140
Devin入門と最近のアップデートから見るDevinの進化 / Introduction to Devin and the Evolution of Devin as Seen in Recent Update
rkaga
9
4.2k
AIコードエディタの基盤となるLLMのFlutter性能評価
alquist4121
0
180
Featured
See All Featured
Save Time (by Creating Custom Rails Generators)
garrettdimon
PRO
30
1.1k
How GitHub (no longer) Works
holman
314
140k
Making Projects Easy
brettharned
116
6.1k
How to Ace a Technical Interview
jacobian
276
23k
Java REST API Framework Comparison - PWX 2021
mraible
29
8.5k
Understanding Cognitive Biases in Performance Measurement
bluesmoon
28
1.6k
Designing for Performance
lara
606
69k
Rails Girls Zürich Keynote
gr2m
94
13k
Bash Introduction
62gerente
611
210k
The Power of CSS Pseudo Elements
geoffreycrofte
75
5.7k
Why You Should Never Use an ORM
jnunemaker
PRO
55
9.3k
It's Worth the Effort
3n
184
28k
Transcript
I S A M I N D S E T
P R O P E R T Y - B A S E D T E S T I N G
@whatyouhide
None
Elixir
T E S T I N G
why do we test? no tests yes tests
unit tests
test "sorting" do assert sort([]) == [] assert sort([1, 2,
3]) == [1, 2, 3] assert sort([2, 1, 3]) == [1, 2, 3] end example-based
table-based Input Output [] [] [1, 2, 3] [1, 2,
3] [2, 1, 3] [1, 2, 3]
unit tests , but also
P R O P E R T I E S
hard to write , but...
valid inputs properties of output testing framework
github.com/whatyouhide/stream_data
example time: sorting lists
test "sorting" do assert sort([]) == [] assert sort([1, 2,
3]) == [1, 2, 3] assert sort([2, 1, 3]) == [1, 2, 3] end
lists of integers
it's a list has the same elements it's ordered
check all list <- list_of(int()) do sorted = sort(list) assert
is_list(sorted) assert same_elements?(list, sorted) assert ordered?(sorted) end
check all list <- list_of(int()) do sorted = sort(list) assert
is_list(sorted) assert same_elements?(list, sorted) assert ordered?(sorted) end
check all list <- list_of(int()) do sorted = sort(list) assert
is_list(sorted) assert same_elements?(list, sorted) assert ordered?(sorted) end
check all list <- list_of(int()) do sorted = sort(list) assert
is_list(sorted) assert same_elements?(list, sorted) assert ordered?(sorted) end
check all list <- list_of(int()) do sorted = sort(list) assert
is_list(sorted) assert same_elements?(list, sorted) assert ordered?(sorted) end
def sort(list), do: list
[32, 2, 44, -12] [1, 0] *shrinking
P A T T E R N S
circular code
decode(encode(term)) == term
property "encoding->decoding is circular" do check all bin <- binary()
do encoded = Huffman.encode(bin) assert is_binary(encoded) assert Huffman.decode(encoded) == bin end end Huffman encoding
oracle model
my_code() == oracle_code()
older system less performant implementation
property "gives same results as Erlang impl" do check all
bin <- binary() do assert Huffman.encode(bin) == :huffman.encode(bin) end end
smoke tests https://www.youtube.com/watch?v=jvwfDdgg93E
API: 200, 201, 400, 404 https://www.youtube.com/watch?v=jvwfDdgg93E
https://www.youtube.com/watch?v=jvwfDdgg93E property "only expected codes are returned" do check all
request <- request() do response = HTTP.perform(request) assert response.status in [200, 201, 400, 404] end end
locally , CI
if System.get_env("CI") == "true" do config :stream_data, max_runs: 500 else
config :stream_data, max_runs: 25 end
unit + properties
R E A L - W O R L D
E X A M P L E
Redis parser
continuation parser parse(bytes) {:ok, command} {:more, function}
property "splitting commands at random" do check all cmd <-
command(), split_cmd <- random_splits(cmd) do assert {:ok, _, ""} = parse_many(split_cmd) end end
property "splitting commands at random" do check all cmd <-
command(), split_cmd <- random_splits(cmd) do assert {:ok, _, ""} = parse_many(split_cmd) end end
property "splitting commands at random" do check all cmd <-
command(), split_cmd <- random_splits(cmd) do assert {:ok, _, ""} = parse_many(split_cmd) end end
S T A T E F U L T E
S T I N G
model valid commands +
model: state + state transformations
commands: calls + preconditions
bounded-size queue
Max size model system
•push(queue, value) •resize(queue, max_size) commands
def push(model, _value), do: model def resize(_model, max_size), do: max_size
push(queue, {:some, "value"}) push(queue, [123, 3.2]) resize(queue, 1) push(queue, %{})
assert size(queue) <= model
LevelDB
17 (seventeen) calls 33 (thirty three) calls
C O N C L U S I O N
find obscure bugs reduce to minimal failing input find specification
errors keep increasing input space
v1.7
use stream_data
use property-based testing
@whatyouhide github.com/whatyouhide/stream_data