/!\ This is a fork of weavejester eftest with changes to make it more t.deps friendly and providing more helpers.
Eftest is a fast and pretty Clojure test runner.
eftest is available on Clojars.
When all the tests pass, it looks like this:
When a test fails, it looks like:
And when a test throws an exception, it looks like:
Eftest has two main functions: find-tests
and run-tests
.
The find-tests
function searches a source, which can be a namespace, directory
path, symbol, var, or a collection of any of the previous. It returns a
collection of test vars found in the source.
The run-tests
function accepts a collection of test vars and runs them,
delivering a report on the tests as it goes.
Typically these two functions are used together:
user=> (require '[s-exp.eftest.runner :refer [find-tests run-tests]])
nil
user=> (run-tests (find-tests "test"))
...
The above example will run all tests found in the "test" directory.
By default Eftest runs all tests in parallel, which can cause issues with tests
that expect to be single-threaded. To disable this and set all tests to be
executed in serial, set the :multithread
option to nil
or #{}
.
user=> (run-tests (find-tests "test") {:multithread nil})
If you want the test vars inside a namespace to be executed in
parallel, but the namespaces themselves to be executed in serial, then
set the :multithread
option to include :vars
:
user=> (run-tests (find-tests "test") {:multithread #{:vars}})
If you want the vars inside a namespace to execute in serial, but the
namespaces to be executed in parallel, set the :multithread
option
to include :namespaces
:
user=> (run-tests (find-tests "test") {:multithread #{:namespaces}})
Alternatively, you can add the :eftest/synchronized
key as metadata to any
tests you want to force to be executed in serial:
(deftest ^:eftest/synchronized a-test
(is (= 1 1)))
Or you can synchronize the entire namespace:
(ns ^:eftest/synchronized foo.core-test
(:require [clojure.test :refer :all]
[foo.core :refer :all]))
When multithreading is enabled, Eftest uses a single fixed-threadpool
ExecutorService
to run all selected tests.
By default, Eftest will instantiate the threadpool with the number of processors
(cores) available to the JVM, as reported by
Runtime.availableProcessors
. (NB: in some
circumstances, such as in a CircleCI test container,
Runtime.availableProcessors
returns an erroneous value.)
Users can override the default behavior by including the key :thread-count
in
the options map supplied to run-tests
with the value being any positive
integer:
user=> (run-tests (find-tests "test") {:thread-count 4})
You can also change the reporting function used. For example, if you want a colorized reporter but without the progress bar:
user=> (run-tests (find-tests "test") {:report s-exp.eftest.report.pretty/report})
Or JUnit output:
user=> (run-tests (find-tests "test") {:report s-exp.eftest.report.junit/report})
Or maybe you just want the old Clojure test reporter:
user=> (run-tests (find-tests "test") {:report clojure.test/report})
If you want to redirect reporting output to a file, use the
s-exp.eftest.report/report-to-file
function:
user=> (require '[s-exp.eftest.report :refer [report-to-file]])
nil
user=&g
68EE
t; (require '[s-exp.eftest.report.junit :as ju])
nil
user=> (run-tests (find-tests "test") {:report (report-to-file ju/report "test.xml")})
By default the STDOUT and STDERR from each test is captured, and displayed only if the test fails. This includes all the output generated by the test itself, and any output generated by other threads not currently running a test.
To turn off this behavior, set the :capture-output
option to
false
:
user=> (run-tests (find-tests "test") {:capture-output false})
Sometimes it's useful to end the testing on the first test failure. To
do this set the :fail-fast
option to true
:
p
user=> (run-tests (find-tests "test") {:fail-fast true})
If you wish to monitor the length of time to run each test, you can
set the :test-warn-time
option to the threshold in milliseconds you
wish to warn on long test for. The measured duration includes the
running of :each-fixtures
, but not :once-fixtures
.
If you know a particular test to be slow and are ok with that, and
don't want to continually be warned about it, you can add the metadata
:eftest/slow
to either the individual test, or the entire namespace,
to prevent reporting.
Note that currently only the pretty
and progress
reporters support
logging long tests.
;; Warns for all tests that take longer than 500ms
user=> (run-tests (find-tests "test") {:test-warn-time 500})
Up to 5a367d5ebd5fef2062b310d7c4602fdc4785edde
Copyright © 2019 James Reeves
Distributed under the Eclipse Public License either version 1.0 or (at your option) any later version.
After 5a367d5ebd5fef2062b310d7c4602fdc4785edde
Copyright © 2025 Max Penet
Distributed under the Eclipse Public License either version 1.0 or (at your option) any later version.