Performance

performance/assoc-many

Enabled by default	Version Added	Version Updated
false	1.10.0	1.10.0

Assoc takes multiple pairs but relies on seq stepping. This is slower than relying on multiple assoc invocations.

Examples

# bad
(assoc m :k1 1 :k2 2 :k3 3)

# good
(-> m
    (assoc :k1 1)
    (assoc :k2 2)
    (assoc :k3 3))

---

performance/avoid-satisfies

Enabled by default	Version Added	Version Updated
false	1.10.0	1.10.0

Avoid use of satisfies? as it is extremely slow. Restructure your code to rely on protocols or other polymorphic forms.

Examples

# bad
(satisfies? Foo :bar)

Reference

• https://bsless.github.io/datahike-datalog-parser

---

performance/dot-equals

Enabled by default	Version Added	Version Updated
false	1.11	1.11

= is quite generalizable and built to handle immutable data. When using a literal, it can be significantly faster to use the underlying Java method.

Currently only checks string literals.

Examples

# bad
(= "foo" s)

# good
(.equals "foo" s)

---

performance/get-in-literals

Enabled by default	Version Added	Version Updated
false	1.10.0	1.10.0

clojure.core/get-in is both polymorphic and relies on seq stepping, which has heavy overhead when the listed slots are keyword literals. Faster to call them as functions.

Examples

# bad
(get-in m [:some-key1 :some-key2 :some-key3])

# good
(-> m :some-key1 :some-key2 :some-key3)

---

performance/get-keyword

Enabled by default	Version Added	Version Updated
false	1.10.0	1.10.0

clojure.core/get is polymorphic and overkill if accessing a map with a keyword literal. The fastest is to fall the map itself as a function but that requires a nil check, so the safest fast method is to use the keyword as function.

Examples

# bad
(get m :some-key)

# good
(:some-key m)

---

performance/into-transducer

Enabled by default	Version Added	Version Updated
false	1.11	1.11

into has a 3-arity and a 4-arity form. Both pour the given coll into the new coll but when given a transducer in the 4-arity form, the transducer is efficiently applied in between.

Examples

# bad
(into [] (map inc (range 100)))

# good
(into [] (map inc) (range 100))

Reference

• https://bsless.github.io/code-smells

---

performance/single-literal-merge

Enabled by default	Version Added	Version Updated
false	1.11	1.11

clojure.core/merge is inherently slow. Its major benefit is handling nil values. If there is only a single object to merge in and it's a map literal, that benefit is doubly unused. Better to directly assoc the values in.

NOTE: If the chosen style is :single and performance/assoc-many is enabled, the style will be treated as :multiple to make the warnings consistent.

Examples

# bad
(merge m {:a 1 :b 2 :c 3})

# good
(assoc m :a 1 :b 2 :c 3)

Configurable Attributes

Name	Default	Options
:chosen-style	:single	:single, :multiple