diff --git a/packaging/hydra.nix b/packaging/hydra.nix
index cef740fa1179..2d86a141e75d 100644
--- a/packaging/hydra.nix
+++ b/packaging/hydra.nix
@@ -440,12 +440,8 @@ rec {
     in
     pkgs.runCommand "install-tests" {
       againstSelf = testNixVersions pkgs pkgs.nix;
-      againstCurrentLatest =
-        # FIXME: temporarily disable this on macOS because of #3605.
-        if system == "x86_64-linux" then testNixVersions pkgs pkgs.nixVersions.latest else null;
-      # Disabled because the latest stable version doesn't handle
-      # `NIX_DAEMON_SOCKET_PATH` which is required for the tests to work
-      # againstLatestStable = testNixVersions pkgs pkgs.nixStable;
+      againstCurrentLatest = testNixVersions pkgs pkgs.nixVersions.latest;
+      againstLatestStable = testNixVersions pkgs pkgs.nixVersions.stable;
     } "touch $out"
   );
 
diff --git a/tests/functional/common/characterisation/framework.sh b/tests/functional/common/characterisation/framework.sh
index b33c165fd2c3..1235b9179176 100644
--- a/tests/functional/common/characterisation/framework.sh
+++ b/tests/functional/common/characterisation/framework.sh
@@ -2,6 +2,19 @@
 
 badTestNames=()
 
+function reportDiffAcceptFailure() {
+    local -r testName=$1
+    # We receive the diff unified stdout/stderr output as the argument
+    printf '%s' "$2" >&2
+    diffExitCode=$?
+    echo >&2 "subtest '$testName' failed: characterised output of $testName not as expected"
+}
+
+function reportDiffAcceptSuccess() {
+    # Do nothing by default
+    :
+}
+
 # Golden test support
 #
 # Test that the output of the given test matches what is expected. If
@@ -16,15 +29,22 @@ function diffAndAcceptInner() {
     if test -e "$expected"; then
         local -r expectedOrEmpty="$expected"
     else
-        local -r expectedOrEmpty=common/characterisation/empty
+        local -r expectedOrEmpty=$_NIX_TEST_SOURCE_DIR/common/characterisation/empty
     fi
 
     # Diff so we get a nice message
-    if ! diff >&2 --color=always --unified "$expectedOrEmpty" "$got"; then
-        echo >&2 "FAIL: evaluation result of $testName not as expected"
+    set +e
+    diffOutput=$(diff 2>&1 --color=always --unified "$expectedOrEmpty" "$got")
+    diffExitCode=$?
+    set -e
+
+    if ((diffExitCode != 0)); then
+        reportDiffAcceptFailure "$diffOutput" "$testName"
         # shellcheck disable=SC2034
         badDiff=1
         badTestNames+=("$testName")
+    else
+        reportDiffAcceptSuccess
     fi
 
     # Update expected if `_NIX_TEST_ACCEPT` is `1`. (Comparing against
diff --git a/tests/functional/common/functions.sh b/tests/functional/common/functions.sh
index cbeb1daaa6d9..54f52ea28486 100644
--- a/tests/functional/common/functions.sh
+++ b/tests/functional/common/functions.sh
@@ -374,7 +374,7 @@ initGitRepo() {
     local extraArgs="${2-}"
 
     # shellcheck disable=SC2086 # word splitting of extraArgs is intended
-    git -C "$repo" init $extraArgs
+    git -C "$repo" init --initial-branch=master $extraArgs
     git -C "$repo" config user.email "foobar@example.com"
     git -C "$repo" config user.name "Foobar"
 }
diff --git a/tests/functional/git/packed-refs-no-cache.sh b/tests/functional/git/packed-refs-no-cache.sh
index 33ffd5fa33e9..fd369e6b1e44 100644
--- a/tests/functional/git/packed-refs-no-cache.sh
+++ b/tests/functional/git/packed-refs-no-cache.sh
@@ -21,11 +21,7 @@ repo=$TEST_ROOT/./git
 
 export _NIX_FORCE_HTTP=1
 
-rm -rf "$repo" "${repo}-tmp" "$TEST_HOME/.cache/nix"
-
-git init --initial-branch="master" "$repo"
-git -C "$repo" config user.email "nix-tests@example.com"
-git -C "$repo" config user.name "Nix Tests"
+createGitRepo "$repo"
 
 echo "hello world" > "$repo/hello_world"
 git -C "$repo" add hello_world
diff --git a/tests/functional/meson.build b/tests/functional/meson.build
index b3f57f64db53..93a50dc9e3fa 100644
--- a/tests/functional/meson.build
+++ b/tests/functional/meson.build
@@ -7,7 +7,7 @@ project(
     'warning_level=1',
     'errorlogs=true', # Please print logs for tests that fail
   ],
-  meson_version : '>= 1.3',
+  meson_version : '>= 1.8',
   license : 'LGPL-2.1-or-later',
 )
 
@@ -206,27 +206,8 @@ if nix_store.found()
 
 endif
 
-# Plugin tests require shared libraries support.
-nix_expr = dependency('nix-expr', required : false)
-if nix_expr.found() and get_option('default_library') != 'static'
-  add_languages('cpp')
-  subdir('plugins')
-  plugin_test_deps = [ libplugintest ]
-  if get_option('plugin-c-api') and not host_machine.system().startswith(
-    'windows',
-  )
-    plugin_test_deps += [ libplugintest_dlsym ]
-  endif
-  suites += {
-    'name' : 'plugins',
-    'deps' : plugin_test_deps,
-    'tests' : [
-      'plugins.sh',
-    ],
-    'workdir' : meson.current_source_dir(),
-  }
-endif
-
+subdir('suites')
+# TODO: Lay out the tests suites in a less flat manner.
 subdir('ca')
 subdir('dyn-drv')
 subdir('flakes')
@@ -260,6 +241,9 @@ foreach suite : suites
       meson.current_build_dir() / 'asan-log',
     )
 
+    overrides = suite.get('overrides', {}).get(script, {})
+    protocol = overrides.get('protocol', 'exitcode')
+
     test(
       name,
       bash,
@@ -280,7 +264,7 @@ foreach suite : suites
         'TEST_NAME' : name,
         'NIX_REMOTE' : '',
         'PS4' : '+(${BASH_SOURCE[0]-$0}:$LINENO) ',
-      },
+      } + suite.get('env', {}),
       # Some tests take 15+ seconds even on an otherwise idle machine;
       # on a loaded machine this can easily drive them to failure. Give
       # them more time than the default of 30 seconds.
@@ -288,6 +272,7 @@ foreach suite : suites
       # Used for target dependency/ordering tracking, not adding compiler flags or anything.
       depends : deps,
       workdir : workdir,
+      protocol : protocol,
     )
   endforeach
 endforeach
diff --git a/tests/functional/plugins.sh b/tests/functional/plugins.sh
deleted file mode 100755
index 948ff1f77eb2..000000000000
--- a/tests/functional/plugins.sh
+++ /dev/null
@@ -1,27 +0,0 @@
-#!/usr/bin/env bash
-
-source common.sh
-
-findPlugin() {
-    local name="$1"
-
-    for ext in so dylib; do
-        local plugin="${_NIX_TEST_BUILD_DIR}/plugins/${name}.$ext"
-        if [[ -f "$plugin" ]]; then
-            printf '%s\n' "$plugin"
-            return 0
-        fi
-    done
-
-    return 1
-}
-
-plugin=$(findPlugin libplugintest)
-
-res=$(nix --option setting-set true --option plugin-files "$plugin" eval --expr builtins.anotherNull)
-
-[ "$res"x = "nullx" ]
-
-if plugin_dlsym=$(findPlugin libplugintest_dlsym); then
-    nix --option plugin-files "$plugin_dlsym" eval --expr 0 >/dev/null
-fi
diff --git a/tests/functional/plugins/meson.build b/tests/functional/plugins/meson.build
deleted file mode 100644
index 4efc9b2a7e8f..000000000000
--- a/tests/functional/plugins/meson.build
+++ /dev/null
@@ -1,20 +0,0 @@
-cpp = meson.get_compiler('cpp')
-dl_dep = cpp.find_library('dl', required : false)
-
-libplugintest = shared_module(
-  'plugintest',
-  'plugintest.cc',
-  dependencies : [
-    dependency('nix-expr'),
-  ],
-  build_by_default : false,
-)
-
-if get_option('plugin-c-api') and not host_machine.system().startswith('windows')
-  libplugintest_dlsym = shared_module(
-    'plugintest_dlsym',
-    'plugintest_dlsym.cc',
-    dependencies : [ dl_dep ],
-    build_by_default : false,
-  )
-endif
diff --git a/tests/functional/repl.sh b/tests/functional/repl.sh
index d63ca76b4e3a..cabb820c9b9e 100755
--- a/tests/functional/repl.sh
+++ b/tests/functional/repl.sh
@@ -1,7 +1,6 @@
 #!/usr/bin/env bash
 
 source common.sh
-source common/characterisation/framework.sh
 
 testDir="$PWD"
 cd "$TEST_ROOT"
@@ -25,13 +24,6 @@ import $testDir/undefined-variable.nix
 
 TODO_NixOS
 
-# FIXME: repl tests fail on systems with stack limits
-stack_ulimit="$(ulimit -Hs)"
-stack_required="$((64 * 1024 * 1024))"
-if [[ "$stack_ulimit" != "unlimited" ]]; then
-    ((stack_ulimit < stack_required)) && skipTest "repl tests cannot run on systems with stack size <$stack_required ($stack_ulimit)"
-fi
-
 testRepl () {
     local nixArgs
     nixArgs=("$@")
@@ -79,270 +71,3 @@ testRepl () {
 testRepl
 # Same thing (kind-of), but with a remote store.
 testRepl --store "$TEST_ROOT/other-root?real=$NIX_STORE_DIR"
-
-# Remove ANSI escape sequences. They can prevent grep from finding a match.
-stripColors () {
-    sed -E 's/\x1B\[([0-9]{1,2}(;[0-9]{1,2})?)?[m|K]//g'
-}
-
-testReplResponseGeneral () {
-    local grepMode commands expectedResponse response
-    grepMode="$1"; shift
-    commands="$1"; shift
-    # Expected response can contain newlines.
-    # grep can't handle multiline patterns, so replace newlines with TEST_NEWLINE
-    # in both expectedResponse and response.
-    # awk ORS always adds a trailing record separator, so we strip it with sed.
-    expectedResponse="$(printf '%s' "$1" | awk 1 ORS=TEST_NEWLINE | sed 's/TEST_NEWLINE$//')"; shift
-    # We don't need to strip trailing record separator here, since extra data is ok.
-    response="$(nix repl "$@" <<< "$commands" 2>&1 | stripColors | awk 1 ORS=TEST_NEWLINE)"
-    printf '%s' "$response" | grepQuiet "$grepMode" -s "$expectedResponse" \
-      || fail "$(echo "repl command set:
-
-$commands
-
-does not respond with:
-
----
-$expectedResponse
----
-
-but with:
-
----
-$response
----
-
-" | sed 's/TEST_NEWLINE/\n/g')"
-}
-
-testReplResponse () {
-    testReplResponseGeneral --basic-regexp "$@"
-}
-
-testReplResponseNoRegex () {
-    testReplResponseGeneral --fixed-strings "$@"
-}
-
-testReplResponse '
-drvPath
-' '".*-simple.drv"' \
---file "$testDir/simple.nix"
-
-testReplResponse '
-drvPath
-' '".*-simple.drv"' \
---file "$testDir/simple.nix" --experimental-features 'ca-derivations'
-
-# `--file` autocalls, `--expr` does not.
-testReplResponse '
-drvPath
-' '".*-simple.drv"' \
---file "$testDir/repl/function.nix"
-
-testReplResponse '
-' 'while evaluating an attribute set to be merged in the global scope' \
---expr '{ x ? 1 }: import '"$testDir"'/simple.nix'
-
-mkdir -p flake && cat <<EOF > flake/flake.nix
-{
-    outputs = { self }: {
-        foo = 1;
-        bar.baz = 2;
-
-        changingThing = "beforeChange";
-    };
-}
-EOF
-testReplResponse '
-foo + baz
-' "3" \
-    ./flake ./flake\#bar --experimental-features 'flakes'
-
-# Test the `:reload` mechanism with flakes:
-# - Eval `./flake#changingThing`
-# - Modify the flake
-# - Re-eval it
-# - Check that the result has changed
-mkfifo repl_fifo
-touch repl_output
-nix repl ./flake --experimental-features 'flakes' < repl_fifo >> repl_output 2>&1 &
-repl_pid=$!
-exec 3>repl_fifo # Open fifo for writing
-echo "changingThing" >&3
-for i in $(seq 1 1000); do
-    if grep -q "beforeChange" repl_output; then
-        break
-    fi
-    cat repl_output
-    sleep 0.1
-done
-if [[ "$i" -eq 100 ]]; then
-    echo "Timed out waiting for beforeChange"
-    exit 1
-fi
-
-sed -i 's/beforeChange/afterChange/' flake/flake.nix
-
-# Send reload and second command
-echo ":reload" >&3
-echo "changingThing" >&3
-echo "exit" >&3
-exec 3>&- # Close fifo
-wait $repl_pid # Wait for process to finish
-grep -q "afterChange" repl_output
-
-# Regression: `:reload` on a flake loaded from a *git* work tree must pick up
-# uncommitted changes. Guards against the per-process workdir-info cache
-# pinning the tree to the rev seen on first load.
-if [[ $(type -p git) ]]; then
-    createGitRepo gitflake
-    cat > gitflake/flake.nix <<EOF
-{ outputs = { self }: { changingThing = "beforeChange"; }; }
-EOF
-    git -C gitflake add flake.nix
-    git -C gitflake commit -m init
-
-    rm -f repl_fifo repl_output
-    mkfifo repl_fifo
-    touch repl_output
-    nix repl ./gitflake --experimental-features 'flakes' < repl_fifo >> repl_output 2>&1 &
-    repl_pid=$!
-    exec 3>repl_fifo
-    echo "changingThing" >&3
-    for _ in $(seq 1 1000); do
-        grep -q "beforeChange" repl_output && break
-        sleep 0.1
-    done
-    grep -q "beforeChange" repl_output || fail "git flake didn't load"
-    sed -i 's/beforeChange/afterChange/' gitflake/flake.nix
-    echo ":reload" >&3
-    echo "changingThing" >&3
-    echo "exit" >&3
-    exec 3>&-
-    wait $repl_pid
-    grep -q "afterChange" repl_output || fail ":reload didn't pick up git work tree change"
-fi
-
-# Regression: a failed `:lf` must not be remembered for `:reload`,
-# and an error in one loaded file must not drop later ones from the reload list.
-testReplResponseNoRegex '
-:lf ./does-not-exist-flake
-:lf ./flake
-:r
-foo
-' '1' \
-    --experimental-features 'flakes'
-
-# Don't prompt for more input when getting unexpected EOF in imported files.
-testReplResponse "
-import $testDir/lang/parse-fail-eof-pos.nix
-" \
-'.*error: syntax error, unexpected end of file.*'
-
-EDITOR='cat' nix repl <<< ':e derivation' 2>&1 | grepQuiet 'derivationStrict'
-EDITOR='cat' nix repl <<< ':e <nix/fetchurl.nix>' 2>&1 | grepQuiet 'builtin:fetchurl'
-
-# TODO: move init to characterisation/framework.sh
-badDiff=0
-badExitCode=0
-
-nixVersion="$(nix eval --impure --raw --expr 'builtins.nixVersion' --extra-experimental-features nix-command)"
-
-# TODO: write a repl interacter for testing. Papering over the differences between readline / editline and between platforms is a pain.
-
-# I couldn't get readline and editline to agree on the newline before the prompt,
-# so let's just force it to be one empty line.
-stripEmptyLinesBeforePrompt() {
-  # --null-data:  treat input as NUL-terminated instead of newline-terminated
-  sed --null-data 's/\n\n*nix-repl>/\n\nnix-repl>/g'
-}
-
-# We don't get a final prompt on darwin, so we strip this as well.
-stripFinalPrompt() {
-  # Strip the final prompt and/or any trailing spaces
-  sed --null-data \
-    -e 's/\(.*[^\n]\)\n\n*nix-repl>[ \n]*$/\1/' \
-    -e 's/[ \n]*$/\n/'
-}
-
-filterReplOutput () {
-  # That is right, we are also filtering out the testdir _without underscores_.
-  # This is crazy, but without it, GHA will fail to run the tests, showing paths
-  # _with_ underscores in the set -x log, but _without_ underscores in the
-  # supposed nix repl output. I have looked in a number of places, but I cannot
-  # find a mechanism that could cause this to happen.
-  local testDirNoUnderscores
-  testDirNoUnderscores="${testDir//_/}"
-
-  stripColors \
-    | tr -d '\0' \
-    | stripEmptyLinesBeforePrompt \
-    | stripFinalPrompt \
-    | sed \
-      -e "s@$testDir@/path/to/tests/functional@g" \
-      -e "s@$testDirNoUnderscores@/path/to/tests/functional@g" \
-      -e "s@$nixVersion@<nix version>@g" \
-    | grep -vF $'warning: you don\'t have Internet access; disabling some network-dependent features' \
-    ;
-}
-
-runRepl () {
-    _NIX_TEST_RAW_MARKDOWN=1 \
-    _NIX_TEST_REPL_ECHO=1 \
-    nix repl "$@" 2>&1 | filterReplOutput
-}
-
-runDebugRepl () {
-    _NIX_TEST_RAW_MARKDOWN=1 \
-    _NIX_TEST_REPL_ECHO=1 \
-    nix eval --file "$1" --debugger "${@:2}" 2>&1 | filterReplOutput
-}
-
-for test in $(cd "$testDir/repl"; echo *.in); do
-    test="$(basename "$test" .in)"
-    in="$testDir/repl/$test.in"
-    actual="$TEST_ROOT/$test.actual"
-    expected="$testDir/repl/$test.expected"
-    declare -a flags=()
-    if test -e "$testDir/repl/$test.flags"; then
-      read -r -a flags < "$testDir/repl/$test.flags"
-    fi
-
-    # Allow putting comments (lines starting with `# COM:`) in the test for
-    # documentation purposes. Regular comments are not skipped, since those are
-    # also interpreted by the repl.
-    inputWithoutComments=$(grep -Ev '^[[:space:]]*#[[:space:]]*COM:' "$in")
-
-    if [ -f "$testDir/repl/$test.in.debugexpr.nix" ]; then
-        if [[ "$test" =~ debugger-fail-.* ]]; then
-            expectedFail=1
-        elif [[ "$test" =~ debugger-okay-.* ]]; then
-            expectedFail=0
-        else
-            die "unexpected debugger test name: '$test', should be either debugger-okay-* or debugger-fail-*"
-        fi
-
-        set +e
-        (cd "$testDir/repl"; set +x; echo "$inputWithoutComments" | runDebugRepl "$testDir/repl/$test.in.debugexpr.nix" "${flags[@]}" 2>&1) > "$actual"
-        debuggerTestCode=$?
-        set -e
-
-        if ((expectedFail == 0 && debuggerTestCode != 0)); then
-            echo "test failed: $test (exit code $debuggerTestCode)" >&2
-            badExitCode=1
-        elif ((expectedFail == 1 && debuggerTestCode == 0)); then
-            echo "test failed: $test unexpectedly succeeded" >&2
-            badExitCode=1
-        fi
-    else
-        (cd "$testDir/repl"; set +x; echo "$inputWithoutComments" | runRepl "${flags[@]}" 2>&1) > "$actual" || {
-            echo "test failed: $test (exit code $?)" >&2
-            badExitCode=1
-        }
-    fi
-
-    diffAndAcceptInner "$test" "$actual" "$expected"
-done
-
-characterisationTestExit
diff --git a/tests/functional/repl/characterisation/empty b/tests/functional/repl/characterisation/empty
deleted file mode 100644
index e69de29bb2d1..000000000000
diff --git a/tests/functional/repl/function.nix b/tests/functional/repl/function.nix
deleted file mode 100644
index 6f9b6071e514..000000000000
--- a/tests/functional/repl/function.nix
+++ /dev/null
@@ -1,4 +0,0 @@
-{
-  x ? 1,
-}:
-import ../simple.nix
diff --git a/tests/functional/suites/meson.build b/tests/functional/suites/meson.build
new file mode 100644
index 000000000000..e59a9e606a70
--- /dev/null
+++ b/tests/functional/suites/meson.build
@@ -0,0 +1,7 @@
+nix_expr = dependency('nix-expr', required : false)
+
+if nix_expr.found() and get_option('default_library') != 'static'
+  subdir('plugins')
+endif
+
+subdir('repl')
diff --git a/tests/functional/suites/plugins/common.sh.in b/tests/functional/suites/plugins/common.sh.in
new file mode 100644
index 000000000000..8bc40751c73b
--- /dev/null
+++ b/tests/functional/suites/plugins/common.sh.in
@@ -0,0 +1,21 @@
+set +x
+
+source "$_NIX_TEST_SOURCE_DIR"/common.sh
+
+set -x
+
+findPlugin() {
+    set +x
+
+    local name="$1"
+
+    for ext in so dylib; do
+        local plugin="@PLUGINS_BUILD_DIR@/${name}.$ext"
+        if [[ -f "$plugin" ]]; then
+            printf '%s\n' "$plugin"
+            return 0
+        fi
+    done
+
+    return 1
+}
diff --git a/tests/functional/suites/plugins/export-c-api.sh b/tests/functional/suites/plugins/export-c-api.sh
new file mode 100644
index 000000000000..1c384b0ce371
--- /dev/null
+++ b/tests/functional/suites/plugins/export-c-api.sh
@@ -0,0 +1,6 @@
+# shellcheck shell=bash
+
+# Test whether -c library API symbols are exported and can be looked up via
+# dlsym.
+
+nix --option plugin-files "$(findPlugin libplugintest_dlsym)" eval --expr 0 >/dev/null
diff --git a/tests/functional/suites/plugins/meson.build b/tests/functional/suites/plugins/meson.build
new file mode 100644
index 000000000000..dd1dc374a8d1
--- /dev/null
+++ b/tests/functional/suites/plugins/meson.build
@@ -0,0 +1,47 @@
+add_languages('cpp', native : false)
+cpp = meson.get_compiler('cpp')
+dl_dep = cpp.find_library('dl', required : false)
+
+plugin_test_deps = []
+plugin_tests = []
+
+libplugintest = shared_module(
+  'plugintest',
+  'src/plugintest.cc',
+  dependencies : [
+    dependency('nix-expr'),
+  ],
+  build_by_default : false,
+)
+
+plugin_tests += [ 'register-setting-and-primop.sh' ]
+plugin_test_deps += [ libplugintest ]
+
+if get_option('plugin-c-api') and not host_machine.system().startswith('windows')
+  libplugintest_dlsym = shared_module(
+    'plugintest_dlsym',
+    'src/plugintest_dlsym.cc',
+    dependencies : [ dl_dep ],
+    build_by_default : false,
+  )
+
+  plugin_tests += [ 'export-c-api.sh' ]
+  plugin_test_deps += [ libplugintest_dlsym ]
+endif
+
+plugin_test_conf = configuration_data()
+plugin_test_conf.set('PLUGINS_BUILD_DIR', meson.current_build_dir())
+
+plugin_test_common = configure_file(
+  input : 'common.sh.in',
+  output : 'common.sh',
+  configuration : plugin_test_conf,
+)
+
+suites += {
+  'name' : 'plugins',
+  'deps' : plugin_test_deps,
+  'tests' : plugin_tests,
+  'workdir' : meson.current_source_dir(),
+  'env' : {'BASH_ENV' : plugin_test_common.full_path()},
+}
diff --git a/tests/functional/suites/plugins/register-setting-and-primop.sh b/tests/functional/suites/plugins/register-setting-and-primop.sh
new file mode 100644
index 000000000000..09315926524f
--- /dev/null
+++ b/tests/functional/suites/plugins/register-setting-and-primop.sh
@@ -0,0 +1,8 @@
+# shellcheck shell=bash
+
+# The loaded plugin registers a primop that behaves differently depending on a
+# setting value.
+
+plugin=$(findPlugin libplugintest)
+res=$(nix --option setting-set true --option plugin-files "$plugin" eval --expr builtins.anotherNull)
+[ "$res"x = "nullx" ]
diff --git a/tests/functional/plugins/plugintest.cc b/tests/functional/suites/plugins/src/plugintest.cc
similarity index 100%
rename from tests/functional/plugins/plugintest.cc
rename to tests/functional/suites/plugins/src/plugintest.cc
diff --git a/tests/functional/plugins/plugintest_dlsym.cc b/tests/functional/suites/plugins/src/plugintest_dlsym.cc
similarity index 100%
rename from tests/functional/plugins/plugintest_dlsym.cc
rename to tests/functional/suites/plugins/src/plugintest_dlsym.cc
diff --git a/tests/functional/repl/add-lots-of-variables.expected b/tests/functional/suites/repl/cases/add-lots-of-variables.expected
similarity index 100%
rename from tests/functional/repl/add-lots-of-variables.expected
rename to tests/functional/suites/repl/cases/add-lots-of-variables.expected
diff --git a/tests/functional/repl/add-lots-of-variables.in b/tests/functional/suites/repl/cases/add-lots-of-variables.in
similarity index 100%
rename from tests/functional/repl/add-lots-of-variables.in
rename to tests/functional/suites/repl/cases/add-lots-of-variables.in
diff --git a/tests/functional/repl/add-overwritten-symbol-env.expected b/tests/functional/suites/repl/cases/add-overwritten-symbol-env.expected
similarity index 100%
rename from tests/functional/repl/add-overwritten-symbol-env.expected
rename to tests/functional/suites/repl/cases/add-overwritten-symbol-env.expected
diff --git a/tests/functional/repl/add-overwritten-symbol-env.in b/tests/functional/suites/repl/cases/add-overwritten-symbol-env.in
similarity index 100%
rename from tests/functional/repl/add-overwritten-symbol-env.in
rename to tests/functional/suites/repl/cases/add-overwritten-symbol-env.in
diff --git a/tests/functional/repl/add-variable-with-spaces.expected b/tests/functional/suites/repl/cases/add-variable-with-spaces.expected
similarity index 100%
rename from tests/functional/repl/add-variable-with-spaces.expected
rename to tests/functional/suites/repl/cases/add-variable-with-spaces.expected
diff --git a/tests/functional/repl/add-variable-with-spaces.in b/tests/functional/suites/repl/cases/add-variable-with-spaces.in
similarity index 100%
rename from tests/functional/repl/add-variable-with-spaces.in
rename to tests/functional/suites/repl/cases/add-variable-with-spaces.in
diff --git a/tests/functional/repl/attribute-set.nix b/tests/functional/suites/repl/cases/attribute-set.nix
similarity index 100%
rename from tests/functional/repl/attribute-set.nix
rename to tests/functional/suites/repl/cases/attribute-set.nix
diff --git a/tests/functional/repl/debugger-fail-throw-env-1.expected b/tests/functional/suites/repl/cases/debugger-fail-throw-env-1.expected
similarity index 64%
rename from tests/functional/repl/debugger-fail-throw-env-1.expected
rename to tests/functional/suites/repl/cases/debugger-fail-throw-env-1.expected
index 921395b8f928..7926a13b1306 100644
--- a/tests/functional/repl/debugger-fail-throw-env-1.expected
+++ b/tests/functional/suites/repl/cases/debugger-fail-throw-env-1.expected
@@ -12,22 +12,22 @@ abort baseNameOf break builtins derivation derivationStrict dirOf false fetchGit
 
 nix-repl> :quit
 error:
-       … while evaluating the file '/path/to/tests/functional/repl/debugger-fail-throw-env-1.in.debugexpr.nix':
+       … while evaluating the file '/path/to/repl/suite/cases/debugger-fail-throw-env-1.in.debugexpr.nix':
 
        … from call site
-         at /path/to/tests/functional/repl/debugger-fail-throw-env-1.in.debugexpr.nix:1:1:
+         at /path/to/repl/suite/cases/debugger-fail-throw-env-1.in.debugexpr.nix:1:1:
             1| (_: throw "oh snap") 42
              | ^
             2|
 
        … while calling anonymous lambda
-         at /path/to/tests/functional/repl/debugger-fail-throw-env-1.in.debugexpr.nix:1:2:
+         at /path/to/repl/suite/cases/debugger-fail-throw-env-1.in.debugexpr.nix:1:2:
             1| (_: throw "oh snap") 42
              |  ^
             2|
 
        … while calling the 'throw' builtin
-         at /path/to/tests/functional/repl/debugger-fail-throw-env-1.in.debugexpr.nix:1:5:
+         at /path/to/repl/suite/cases/debugger-fail-throw-env-1.in.debugexpr.nix:1:5:
             1| (_: throw "oh snap") 42
              |     ^
             2|
diff --git a/tests/functional/repl/debugger-fail-throw-env-1.in b/tests/functional/suites/repl/cases/debugger-fail-throw-env-1.in
similarity index 100%
rename from tests/functional/repl/debugger-fail-throw-env-1.in
rename to tests/functional/suites/repl/cases/debugger-fail-throw-env-1.in
diff --git a/tests/functional/repl/debugger-fail-throw-env-1.in.debugexpr.nix b/tests/functional/suites/repl/cases/debugger-fail-throw-env-1.in.debugexpr.nix
similarity index 100%
rename from tests/functional/repl/debugger-fail-throw-env-1.in.debugexpr.nix
rename to tests/functional/suites/repl/cases/debugger-fail-throw-env-1.in.debugexpr.nix
diff --git a/tests/functional/repl/debugger-fail-throw-env-2.expected b/tests/functional/suites/repl/cases/debugger-fail-throw-env-2.expected
similarity index 64%
rename from tests/functional/repl/debugger-fail-throw-env-2.expected
rename to tests/functional/suites/repl/cases/debugger-fail-throw-env-2.expected
index 3eb47bdf53aa..e2f9d5ac47ba 100644
--- a/tests/functional/repl/debugger-fail-throw-env-2.expected
+++ b/tests/functional/suites/repl/cases/debugger-fail-throw-env-2.expected
@@ -20,7 +20,7 @@ abort baseNameOf break builtins derivation derivationStrict dirOf false fetchGit
 nix-repl> :bt
 
 0: error: oh snap
-/path/to/tests/functional/repl/debugger-fail-throw-env-2.in.debugexpr.nix:5:23
+/path/to/repl/suite/cases/debugger-fail-throw-env-2.in.debugexpr.nix:5:23
 
      4| with x;
      5| (_: builtins.seq x.a (throw "oh snap")) x.a
@@ -28,7 +28,7 @@ nix-repl> :bt
      6| 
 
 1: while calling a function
-/path/to/tests/functional/repl/debugger-fail-throw-env-2.in.debugexpr.nix:5:23
+/path/to/repl/suite/cases/debugger-fail-throw-env-2.in.debugexpr.nix:5:23
 
      4| with x;
      5| (_: builtins.seq x.a (throw "oh snap")) x.a
@@ -36,7 +36,7 @@ nix-repl> :bt
      6| 
 
 2: while calling a function
-/path/to/tests/functional/repl/debugger-fail-throw-env-2.in.debugexpr.nix:5:5
+/path/to/repl/suite/cases/debugger-fail-throw-env-2.in.debugexpr.nix:5:5
 
      4| with x;
      5| (_: builtins.seq x.a (throw "oh snap")) x.a
@@ -44,7 +44,7 @@ nix-repl> :bt
      6| 
 
 3: while calling anonymous lambda
-/path/to/tests/functional/repl/debugger-fail-throw-env-2.in.debugexpr.nix:5:2
+/path/to/repl/suite/cases/debugger-fail-throw-env-2.in.debugexpr.nix:5:2
 
      4| with x;
      5| (_: builtins.seq x.a (throw "oh snap")) x.a
@@ -52,7 +52,7 @@ nix-repl> :bt
      6| 
 
 4: while calling a function
-/path/to/tests/functional/repl/debugger-fail-throw-env-2.in.debugexpr.nix:5:1
+/path/to/repl/suite/cases/debugger-fail-throw-env-2.in.debugexpr.nix:5:1
 
      4| with x;
      5| (_: builtins.seq x.a (throw "oh snap")) x.a
@@ -61,35 +61,35 @@ nix-repl> :bt
 
 5: while evaluating a 'let' expression
 
-6: while evaluating the file '/path/to/tests/functional/repl/debugger-fail-throw-env-2.in.debugexpr.nix':
+6: while evaluating the file '/path/to/repl/suite/cases/debugger-fail-throw-env-2.in.debugexpr.nix':
 
 
 error:
-       … while evaluating the file '/path/to/tests/functional/repl/debugger-fail-throw-env-2.in.debugexpr.nix':
+       … while evaluating the file '/path/to/repl/suite/cases/debugger-fail-throw-env-2.in.debugexpr.nix':
 
        … from call site
-         at /path/to/tests/functional/repl/debugger-fail-throw-env-2.in.debugexpr.nix:5:1:
+         at /path/to/repl/suite/cases/debugger-fail-throw-env-2.in.debugexpr.nix:5:1:
             4| with x;
             5| (_: builtins.seq x.a (throw "oh snap")) x.a
              | ^
             6|
 
        … while calling anonymous lambda
-         at /path/to/tests/functional/repl/debugger-fail-throw-env-2.in.debugexpr.nix:5:2:
+         at /path/to/repl/suite/cases/debugger-fail-throw-env-2.in.debugexpr.nix:5:2:
             4| with x;
             5| (_: builtins.seq x.a (throw "oh snap")) x.a
              |  ^
             6|
 
        … while calling the 'seq' builtin
-         at /path/to/tests/functional/repl/debugger-fail-throw-env-2.in.debugexpr.nix:5:5:
+         at /path/to/repl/suite/cases/debugger-fail-throw-env-2.in.debugexpr.nix:5:5:
             4| with x;
             5| (_: builtins.seq x.a (throw "oh snap")) x.a
              |     ^
             6|
 
        … while calling the 'throw' builtin
-         at /path/to/tests/functional/repl/debugger-fail-throw-env-2.in.debugexpr.nix:5:23:
+         at /path/to/repl/suite/cases/debugger-fail-throw-env-2.in.debugexpr.nix:5:23:
             4| with x;
             5| (_: builtins.seq x.a (throw "oh snap")) x.a
              |                       ^
diff --git a/tests/functional/repl/debugger-fail-throw-env-2.in b/tests/functional/suites/repl/cases/debugger-fail-throw-env-2.in
similarity index 100%
rename from tests/functional/repl/debugger-fail-throw-env-2.in
rename to tests/functional/suites/repl/cases/debugger-fail-throw-env-2.in
diff --git a/tests/functional/repl/debugger-fail-throw-env-2.in.debugexpr.nix b/tests/functional/suites/repl/cases/debugger-fail-throw-env-2.in.debugexpr.nix
similarity index 100%
rename from tests/functional/repl/debugger-fail-throw-env-2.in.debugexpr.nix
rename to tests/functional/suites/repl/cases/debugger-fail-throw-env-2.in.debugexpr.nix
diff --git a/tests/functional/repl/debugger-okay-break.expected b/tests/functional/suites/repl/cases/debugger-okay-break.expected
similarity index 59%
rename from tests/functional/repl/debugger-okay-break.expected
rename to tests/functional/suites/repl/cases/debugger-okay-break.expected
index 0cec2398a75e..9d1b094b65b7 100644
--- a/tests/functional/repl/debugger-okay-break.expected
+++ b/tests/functional/suites/repl/cases/debugger-okay-break.expected
@@ -8,7 +8,7 @@ nix-repl> :bt
 0: error: breakpoint reached
 
 1: while calling a function
-/path/to/tests/functional/repl/debugger-okay-break.in.debugexpr.nix:5:1
+/path/to/repl/suite/cases/debugger-okay-break.in.debugexpr.nix:5:1
 
      4| in
      5| builtins.break othervar
@@ -17,14 +17,14 @@ nix-repl> :bt
 
 2: while evaluating a 'let' expression
 
-3: while evaluating the file '/path/to/tests/functional/repl/debugger-okay-break.in.debugexpr.nix':
+3: while evaluating the file '/path/to/repl/suite/cases/debugger-okay-break.in.debugexpr.nix':
 
 nix-repl> :backtrace
 
 0: error: breakpoint reached
 
 1: while calling a function
-/path/to/tests/functional/repl/debugger-okay-break.in.debugexpr.nix:5:1
+/path/to/repl/suite/cases/debugger-okay-break.in.debugexpr.nix:5:1
 
      4| in
      5| builtins.break othervar
@@ -33,7 +33,7 @@ nix-repl> :backtrace
 
 2: while evaluating a 'let' expression
 
-3: while evaluating the file '/path/to/tests/functional/repl/debugger-okay-break.in.debugexpr.nix':
+3: while evaluating the file '/path/to/repl/suite/cases/debugger-okay-break.in.debugexpr.nix':
 
 nix-repl> :p somevar
 hello
diff --git a/tests/functional/repl/debugger-okay-break.in b/tests/functional/suites/repl/cases/debugger-okay-break.in
similarity index 100%
rename from tests/functional/repl/debugger-okay-break.in
rename to tests/functional/suites/repl/cases/debugger-okay-break.in
diff --git a/tests/functional/repl/debugger-okay-break.in.debugexpr.nix b/tests/functional/suites/repl/cases/debugger-okay-break.in.debugexpr.nix
similarity index 100%
rename from tests/functional/repl/debugger-okay-break.in.debugexpr.nix
rename to tests/functional/suites/repl/cases/debugger-okay-break.in.debugexpr.nix
diff --git a/tests/functional/repl/debugger-okay-ignore-try.expected b/tests/functional/suites/repl/cases/debugger-okay-ignore-try.expected
similarity index 90%
rename from tests/functional/repl/debugger-okay-ignore-try.expected
rename to tests/functional/suites/repl/cases/debugger-okay-ignore-try.expected
index 7d4e72fdaeab..1c168f96c47d 100644
--- a/tests/functional/repl/debugger-okay-ignore-try.expected
+++ b/tests/functional/suites/repl/cases/debugger-okay-ignore-try.expected
@@ -8,7 +8,7 @@ nix-repl> :bt
 0: error: breakpoint reached
 
 1: while calling a function
-/path/to/tests/functional/repl/debugger-okay-ignore-try.in.debugexpr.nix:6:13
+/path/to/repl/suite/cases/debugger-okay-ignore-try.in.debugexpr.nix:6:13
 
      5| let
      6|   bricked = break tried;
@@ -19,7 +19,7 @@ nix-repl> :bt
 
 3: while evaluating a 'let' expression
 
-4: while evaluating the file '/path/to/tests/functional/repl/debugger-okay-ignore-try.in.debugexpr.nix':
+4: while evaluating the file '/path/to/repl/suite/cases/debugger-okay-ignore-try.in.debugexpr.nix':
 
 nix-repl> :st 3
 
@@ -71,7 +71,7 @@ nix-repl> :p tried
 nix-repl> :st 1
 
 1: while calling a function
-/path/to/tests/functional/repl/debugger-okay-ignore-try.in.debugexpr.nix:6:13
+/path/to/repl/suite/cases/debugger-okay-ignore-try.in.debugexpr.nix:6:13
 
      5| let
      6|   bricked = break tried;
diff --git a/tests/functional/repl/debugger-okay-ignore-try.flags b/tests/functional/suites/repl/cases/debugger-okay-ignore-try.flags
similarity index 100%
rename from tests/functional/repl/debugger-okay-ignore-try.flags
rename to tests/functional/suites/repl/cases/debugger-okay-ignore-try.flags
diff --git a/tests/functional/repl/debugger-okay-ignore-try.in b/tests/functional/suites/repl/cases/debugger-okay-ignore-try.in
similarity index 100%
rename from tests/functional/repl/debugger-okay-ignore-try.in
rename to tests/functional/suites/repl/cases/debugger-okay-ignore-try.in
diff --git a/tests/functional/repl/debugger-okay-ignore-try.in.debugexpr.nix b/tests/functional/suites/repl/cases/debugger-okay-ignore-try.in.debugexpr.nix
similarity index 100%
rename from tests/functional/repl/debugger-okay-ignore-try.in.debugexpr.nix
rename to tests/functional/suites/repl/cases/debugger-okay-ignore-try.in.debugexpr.nix
diff --git a/tests/functional/repl/debugger-okay-on-warn.expected b/tests/functional/suites/repl/cases/debugger-okay-on-warn.expected
similarity index 80%
rename from tests/functional/repl/debugger-okay-on-warn.expected
rename to tests/functional/suites/repl/cases/debugger-okay-on-warn.expected
index 53fb13112522..37a253df5ddd 100644
--- a/tests/functional/repl/debugger-okay-on-warn.expected
+++ b/tests/functional/suites/repl/cases/debugger-okay-on-warn.expected
@@ -5,7 +5,7 @@ Type :? for help.
 nix-repl> :p fails
 error:
        … while calling the 'throw' builtin
-         at /path/to/tests/functional/repl/debugger-okay-on-warn.in.debugexpr.nix:3:11:
+         at /path/to/repl/suite/cases/debugger-okay-on-warn.in.debugexpr.nix:3:11:
             2|   msg = "hello";
             3|   fails = throw (toString 1);
              |           ^
diff --git a/tests/functional/repl/debugger-okay-on-warn.flags b/tests/functional/suites/repl/cases/debugger-okay-on-warn.flags
similarity index 100%
rename from tests/functional/repl/debugger-okay-on-warn.flags
rename to tests/functional/suites/repl/cases/debugger-okay-on-warn.flags
diff --git a/tests/functional/repl/debugger-okay-on-warn.in b/tests/functional/suites/repl/cases/debugger-okay-on-warn.in
similarity index 100%
rename from tests/functional/repl/debugger-okay-on-warn.in
rename to tests/functional/suites/repl/cases/debugger-okay-on-warn.in
diff --git a/tests/functional/repl/debugger-okay-on-warn.in.debugexpr.nix b/tests/functional/suites/repl/cases/debugger-okay-on-warn.in.debugexpr.nix
similarity index 100%
rename from tests/functional/repl/debugger-okay-on-warn.in.debugexpr.nix
rename to tests/functional/suites/repl/cases/debugger-okay-on-warn.in.debugexpr.nix
diff --git a/tests/functional/repl/doc-comment-curried-args.expected b/tests/functional/suites/repl/cases/doc-comment-curried-args.expected
similarity index 84%
rename from tests/functional/repl/doc-comment-curried-args.expected
rename to tests/functional/suites/repl/cases/doc-comment-curried-args.expected
index 29ca40cee2cc..ab6a8e18f991 100644
--- a/tests/functional/repl/doc-comment-curried-args.expected
+++ b/tests/functional/suites/repl/cases/doc-comment-curried-args.expected
@@ -7,7 +7,7 @@ compact, constant, curriedArgs, documentedFormals, floatedIn, lib, measurement,
 
 nix-repl> :doc curriedArgs
 Function `curriedArgs`\
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:87:5
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:87:5
 
 
 A documented function.
@@ -19,7 +19,7 @@ nix-repl> "Note that users may not expect this to behave as it currently does"
 
 nix-repl> :doc x
 Function `curriedArgs`\
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:91:5
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:91:5
 
 
 The function returned by applying once
diff --git a/tests/functional/repl/doc-comment-curried-args.in b/tests/functional/suites/repl/cases/doc-comment-curried-args.in
similarity index 100%
rename from tests/functional/repl/doc-comment-curried-args.in
rename to tests/functional/suites/repl/cases/doc-comment-curried-args.in
diff --git a/tests/functional/repl/doc-comment-formals.expected b/tests/functional/suites/repl/cases/doc-comment-formals.expected
similarity index 84%
rename from tests/functional/repl/doc-comment-formals.expected
rename to tests/functional/suites/repl/cases/doc-comment-formals.expected
index 65fbc2bd5d80..b94e8763c293 100644
--- a/tests/functional/repl/doc-comment-formals.expected
+++ b/tests/functional/suites/repl/cases/doc-comment-formals.expected
@@ -10,7 +10,7 @@ nix-repl> "Note that this is not yet complete"
 
 nix-repl> :doc documentedFormals
 Function `documentedFormals`\
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:104:5
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:104:5
 
 
 Finds x
diff --git a/tests/functional/repl/doc-comment-formals.in b/tests/functional/suites/repl/cases/doc-comment-formals.in
similarity index 100%
rename from tests/functional/repl/doc-comment-formals.in
rename to tests/functional/suites/repl/cases/doc-comment-formals.in
diff --git a/tests/functional/repl/doc-comment-function.expected b/tests/functional/suites/repl/cases/doc-comment-function.expected
similarity index 64%
rename from tests/functional/repl/doc-comment-function.expected
rename to tests/functional/suites/repl/cases/doc-comment-function.expected
index 030cfc3265a1..84b2c1826a65 100644
--- a/tests/functional/repl/doc-comment-function.expected
+++ b/tests/functional/suites/repl/cases/doc-comment-function.expected
@@ -2,7 +2,7 @@ Nix <nix version>
 Type :? for help.
 
 nix-repl> :doc import ./doc-comment-function.nix
-Function defined at /path/to/tests/functional/repl/doc-comment-function.nix:4:1
+Function defined at /path/to/repl/suite/cases/doc-comment-function.nix:4:1
 
  
 A doc comment for a file that only contains a function
diff --git a/tests/functional/repl/doc-comment-function.in b/tests/functional/suites/repl/cases/doc-comment-function.in
similarity index 100%
rename from tests/functional/repl/doc-comment-function.in
rename to tests/functional/suites/repl/cases/doc-comment-function.in
diff --git a/tests/functional/repl/doc-comment-function.nix b/tests/functional/suites/repl/cases/doc-comment-function.nix
similarity index 100%
rename from tests/functional/repl/doc-comment-function.nix
rename to tests/functional/suites/repl/cases/doc-comment-function.nix
diff --git a/tests/functional/repl/doc-comments.nix b/tests/functional/suites/repl/cases/doc-comments.nix
similarity index 100%
rename from tests/functional/repl/doc-comments.nix
rename to tests/functional/suites/repl/cases/doc-comments.nix
diff --git a/tests/functional/repl/doc-compact.expected b/tests/functional/suites/repl/cases/doc-compact.expected
similarity index 80%
rename from tests/functional/repl/doc-compact.expected
rename to tests/functional/suites/repl/cases/doc-compact.expected
index 9fcf6169a5fc..1a46c65a71c6 100644
--- a/tests/functional/repl/doc-compact.expected
+++ b/tests/functional/suites/repl/cases/doc-compact.expected
@@ -7,7 +7,7 @@ compact, constant, curriedArgs, documentedFormals, floatedIn, lib, measurement,
 
 nix-repl> :doc compact
 Function `compact`\
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:27:5
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:27:5
 
 
 boom
diff --git a/tests/functional/repl/doc-compact.in b/tests/functional/suites/repl/cases/doc-compact.in
similarity index 100%
rename from tests/functional/repl/doc-compact.in
rename to tests/functional/suites/repl/cases/doc-compact.in
diff --git a/tests/functional/repl/doc-constant.expected b/tests/functional/suites/repl/cases/doc-constant.expected
similarity index 81%
rename from tests/functional/repl/doc-constant.expected
rename to tests/functional/suites/repl/cases/doc-constant.expected
index e2ccd0945080..f57ab7ac425e 100644
--- a/tests/functional/repl/doc-constant.expected
+++ b/tests/functional/suites/repl/cases/doc-constant.expected
@@ -11,7 +11,7 @@ error: value does not have documentation
 nix-repl> :doc lib.version
 Attribute `version`
 
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:47:3
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:47:3
 
 
 Immovably fixed.
@@ -19,7 +19,7 @@ Immovably fixed.
 nix-repl> :doc lib.attr.empty
 Attribute `empty`
 
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:52:3
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:52:3
 
 
 Unchangeably constant.
@@ -32,7 +32,7 @@ error:
              | ^
 
        … from the definition of 'attr'
-         at /path/to/tests/functional/repl/doc-comments.nix:52:3:
+         at /path/to/repl/suite/cases/doc-comments.nix:52:3:
            51|   */
            52|   lib.attr.empty = { };
              |   ^
@@ -44,7 +44,7 @@ error:
 nix-repl> :doc (import ./doc-comments.nix).constant
 Attribute `constant`
 
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:42:3
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:42:3
 
 
 Firmly rigid.
@@ -52,7 +52,7 @@ Firmly rigid.
 nix-repl> :doc (import ./doc-comments.nix).lib.version
 Attribute `version`
 
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:47:3
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:47:3
 
 
 Immovably fixed.
@@ -60,7 +60,7 @@ Immovably fixed.
 nix-repl> :doc (import ./doc-comments.nix).lib.attr.empty
 Attribute `empty`
 
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:52:3
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:52:3
 
 
 Unchangeably constant.
@@ -68,7 +68,7 @@ Unchangeably constant.
 nix-repl> :doc (import ./doc-comments.nix).lib.attr.undocumented
 Attribute `undocumented`
 
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:54:3
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:54:3
 
 No documentation found.
 
@@ -116,7 +116,7 @@ error:
              | ^
 
        … from the definition of 'attr'
-         at /path/to/tests/functional/repl/doc-comments.nix:52:3:
+         at /path/to/repl/suite/cases/doc-comments.nix:52:3:
            51|   */
            52|   lib.attr.empty = { };
              |   ^
diff --git a/tests/functional/repl/doc-constant.in b/tests/functional/suites/repl/cases/doc-constant.in
similarity index 100%
rename from tests/functional/repl/doc-constant.in
rename to tests/functional/suites/repl/cases/doc-constant.in
diff --git a/tests/functional/repl/doc-floatedIn.expected b/tests/functional/suites/repl/cases/doc-floatedIn.expected
similarity index 80%
rename from tests/functional/repl/doc-floatedIn.expected
rename to tests/functional/suites/repl/cases/doc-floatedIn.expected
index e01f8868ee7b..8baa7ba37f6c 100644
--- a/tests/functional/repl/doc-floatedIn.expected
+++ b/tests/functional/suites/repl/cases/doc-floatedIn.expected
@@ -7,7 +7,7 @@ compact, constant, curriedArgs, documentedFormals, floatedIn, lib, measurement,
 
 nix-repl> :doc floatedIn
 Function `floatedIn`\
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:21:5
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:21:5
 
 
 This also works.
diff --git a/tests/functional/repl/doc-floatedIn.in b/tests/functional/suites/repl/cases/doc-floatedIn.in
similarity index 100%
rename from tests/functional/repl/doc-floatedIn.in
rename to tests/functional/suites/repl/cases/doc-floatedIn.in
diff --git a/tests/functional/repl/doc-functor.expected b/tests/functional/suites/repl/cases/doc-functor.expected
similarity index 71%
rename from tests/functional/repl/doc-functor.expected
rename to tests/functional/suites/repl/cases/doc-functor.expected
index e5984098c344..0a81c0f67fea 100644
--- a/tests/functional/repl/doc-functor.expected
+++ b/tests/functional/suites/repl/cases/doc-functor.expected
@@ -7,21 +7,21 @@ diverging, doubler, helper, helper2, helper3, lib, makeOverridable, makeVeryOver
 
 nix-repl> :doc multiplier
 Function `__functor`\
-  … defined at /path/to/tests/functional/repl/doc-functor.nix:12:23
+  … defined at /path/to/repl/suite/cases/doc-functor.nix:12:23
 
 
 Multiply the argument by the factor stored in the factor attribute.
 
 nix-repl> :doc doubler
 Function `multiply`\
-  … defined at /path/to/tests/functional/repl/doc-functor.nix:5:17
+  … defined at /path/to/repl/suite/cases/doc-functor.nix:5:17
 
 
 Look, it's just like a function!
 
 nix-repl> :doc recursive
 Function `__functor`\
-  … defined at /path/to/tests/functional/repl/doc-functor.nix:82:23
+  … defined at /path/to/repl/suite/cases/doc-functor.nix:82:23
 
 
 This looks bad, but the docs are ok because of the eta expansion.
@@ -31,14 +31,14 @@ error:
        … while partially calling '__functor' to retrieve documentation
 
        … while calling '__functor'
-         at /path/to/tests/functional/repl/doc-functor.nix:90:17:
+         at /path/to/repl/suite/cases/doc-functor.nix:90:17:
            89|     */
            90|     __functor = self: self.__functor self;
              |                 ^
            91|   };
 
        … from call site
-         at /path/to/tests/functional/repl/doc-functor.nix:90:23:
+         at /path/to/repl/suite/cases/doc-functor.nix:90:23:
            89|     */
            90|     __functor = self: self.__functor self;
              |                       ^
@@ -47,7 +47,7 @@ error:
        (199 duplicate frames omitted)
 
        error: stack overflow; max-call-depth exceeded
-       at /path/to/tests/functional/repl/doc-functor.nix:90:23:
+       at /path/to/repl/suite/cases/doc-functor.nix:90:23:
            89|     */
            90|     __functor = self: self.__functor self;
              |                       ^
@@ -60,14 +60,14 @@ error:
        (100 duplicate frames omitted)
 
        … while calling '__functor'
-         at /path/to/tests/functional/repl/doc-functor.nix:103:21:
+         at /path/to/repl/suite/cases/doc-functor.nix:103:21:
           102|       f = x: {
           103|         __functor = self: (f (x + 1));
              |                     ^
           104|       };
 
        error: stack overflow; max-call-depth exceeded
-       at /path/to/tests/functional/repl/doc-functor.nix:103:28:
+       at /path/to/repl/suite/cases/doc-functor.nix:103:28:
           102|       f = x: {
           103|         __functor = self: (f (x + 1));
              |                            ^
@@ -75,28 +75,28 @@ error:
 
 nix-repl> :doc helper
 Function `square`\
-  … defined at /path/to/tests/functional/repl/doc-functor.nix:36:12
+  … defined at /path/to/repl/suite/cases/doc-functor.nix:36:12
 
 
 Compute x^2
 
 nix-repl> :doc helper2
 Function `__functor`\
-  … defined at /path/to/tests/functional/repl/doc-functor.nix:46:13
+  … defined at /path/to/repl/suite/cases/doc-functor.nix:46:13
 
 
 This is a function that can be overridden.
 
 nix-repl> :doc lib.helper3
 Function `__functor`\
-  … defined at /path/to/tests/functional/repl/doc-functor.nix:46:13
+  … defined at /path/to/repl/suite/cases/doc-functor.nix:46:13
 
 
 This is a function that can be overridden.
 
 nix-repl> :doc helper3
 Function `__functor`\
-  … defined at /path/to/tests/functional/repl/doc-functor.nix:46:13
+  … defined at /path/to/repl/suite/cases/doc-functor.nix:46:13
 
 
 This is a function that can be overridden.
diff --git a/tests/functional/repl/doc-functor.flags b/tests/functional/suites/repl/cases/doc-functor.flags
similarity index 100%
rename from tests/functional/repl/doc-functor.flags
rename to tests/functional/suites/repl/cases/doc-functor.flags
diff --git a/tests/functional/repl/doc-functor.in b/tests/functional/suites/repl/cases/doc-functor.in
similarity index 100%
rename from tests/functional/repl/doc-functor.in
rename to tests/functional/suites/repl/cases/doc-functor.in
diff --git a/tests/functional/repl/doc-functor.nix b/tests/functional/suites/repl/cases/doc-functor.nix
similarity index 100%
rename from tests/functional/repl/doc-functor.nix
rename to tests/functional/suites/repl/cases/doc-functor.nix
diff --git a/tests/functional/repl/doc-lambda-flavors.expected b/tests/functional/suites/repl/cases/doc-lambda-flavors.expected
similarity index 67%
rename from tests/functional/repl/doc-lambda-flavors.expected
rename to tests/functional/suites/repl/cases/doc-lambda-flavors.expected
index 676ef73135dd..d1d30ed1ab3b 100644
--- a/tests/functional/repl/doc-lambda-flavors.expected
+++ b/tests/functional/suites/repl/cases/doc-lambda-flavors.expected
@@ -7,28 +7,28 @@ compact, constant, curriedArgs, documentedFormals, floatedIn, lib, measurement,
 
 nix-repl> :doc nonStrict
 Function `nonStrict`\
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:60:5
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:60:5
 
 
 My syntax is not strict, but I'm strict anyway.
 
 nix-repl> :doc strict
 Function `strict`\
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:65:5
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:65:5
 
 
 I don't have to be strict, but I am anyway.
 
 nix-repl> :doc strictPre
 Function `strictPre`\
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:71:5
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:71:5
 
 
 Here's one way to do this
 
 nix-repl> :doc strictPost
 Function `strictPost`\
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:76:5
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:76:5
 
 
 Here's another way to do this
diff --git a/tests/functional/repl/doc-lambda-flavors.in b/tests/functional/suites/repl/cases/doc-lambda-flavors.in
similarity index 100%
rename from tests/functional/repl/doc-lambda-flavors.in
rename to tests/functional/suites/repl/cases/doc-lambda-flavors.in
diff --git a/tests/functional/repl/doc-measurement.expected b/tests/functional/suites/repl/cases/doc-measurement.expected
similarity index 81%
rename from tests/functional/repl/doc-measurement.expected
rename to tests/functional/suites/repl/cases/doc-measurement.expected
index ef176cea883f..ac670b5e7ae2 100644
--- a/tests/functional/repl/doc-measurement.expected
+++ b/tests/functional/suites/repl/cases/doc-measurement.expected
@@ -7,7 +7,7 @@ compact, constant, curriedArgs, documentedFormals, floatedIn, lib, measurement,
 
 nix-repl> :doc measurement
 Function `measurement`\
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:15:17
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:15:17
 
 
 👈 precisely this wide 👉
diff --git a/tests/functional/repl/doc-measurement.in b/tests/functional/suites/repl/cases/doc-measurement.in
similarity index 100%
rename from tests/functional/repl/doc-measurement.in
rename to tests/functional/suites/repl/cases/doc-measurement.in
diff --git a/tests/functional/repl/doc-multiply.expected b/tests/functional/suites/repl/cases/doc-multiply.expected
similarity index 84%
rename from tests/functional/repl/doc-multiply.expected
rename to tests/functional/suites/repl/cases/doc-multiply.expected
index c2024b094918..b33e900995d1 100644
--- a/tests/functional/repl/doc-multiply.expected
+++ b/tests/functional/suites/repl/cases/doc-multiply.expected
@@ -7,7 +7,7 @@ compact, constant, curriedArgs, documentedFormals, floatedIn, lib, measurement,
 
 nix-repl> :doc multiply
 Function `multiply`\
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:10:14
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:10:14
 
 
 Perform *arithmetic* multiplication. It's kind of like repeated **addition**, very neat.
diff --git a/tests/functional/repl/doc-multiply.in b/tests/functional/suites/repl/cases/doc-multiply.in
similarity index 100%
rename from tests/functional/repl/doc-multiply.in
rename to tests/functional/suites/repl/cases/doc-multiply.in
diff --git a/tests/functional/repl/doc-unambiguous.expected b/tests/functional/suites/repl/cases/doc-unambiguous.expected
similarity index 80%
rename from tests/functional/repl/doc-unambiguous.expected
rename to tests/functional/suites/repl/cases/doc-unambiguous.expected
index d2a452ba7bf3..66e9235b4ede 100644
--- a/tests/functional/repl/doc-unambiguous.expected
+++ b/tests/functional/suites/repl/cases/doc-unambiguous.expected
@@ -7,7 +7,7 @@ compact, constant, curriedArgs, documentedFormals, floatedIn, lib, measurement,
 
 nix-repl> :doc unambiguous
 Function `unambiguous`\
-  … defined at /path/to/tests/functional/repl/doc-comments.nix:37:5
+  … defined at /path/to/repl/suite/cases/doc-comments.nix:37:5
 
 
 Very close
diff --git a/tests/functional/repl/doc-unambiguous.in b/tests/functional/suites/repl/cases/doc-unambiguous.in
similarity index 100%
rename from tests/functional/repl/doc-unambiguous.in
rename to tests/functional/suites/repl/cases/doc-unambiguous.in
diff --git a/tests/functional/repl/dollar-escaping.expected b/tests/functional/suites/repl/cases/dollar-escaping.expected
similarity index 100%
rename from tests/functional/repl/dollar-escaping.expected
rename to tests/functional/suites/repl/cases/dollar-escaping.expected
diff --git a/tests/functional/repl/dollar-escaping.in b/tests/functional/suites/repl/cases/dollar-escaping.in
similarity index 100%
rename from tests/functional/repl/dollar-escaping.in
rename to tests/functional/suites/repl/cases/dollar-escaping.in
diff --git a/tests/functional/repl/file-a.nix b/tests/functional/suites/repl/cases/file-a.nix
similarity index 100%
rename from tests/functional/repl/file-a.nix
rename to tests/functional/suites/repl/cases/file-a.nix
diff --git a/tests/functional/repl/file-b.nix b/tests/functional/suites/repl/cases/file-b.nix
similarity index 100%
rename from tests/functional/repl/file-b.nix
rename to tests/functional/suites/repl/cases/file-b.nix
diff --git a/tests/functional/repl/inherit-and-assignment.expected b/tests/functional/suites/repl/cases/inherit-and-assignment.expected
similarity index 100%
rename from tests/functional/repl/inherit-and-assignment.expected
rename to tests/functional/suites/repl/cases/inherit-and-assignment.expected
diff --git a/tests/functional/repl/inherit-and-assignment.in b/tests/functional/suites/repl/cases/inherit-and-assignment.in
similarity index 100%
rename from tests/functional/repl/inherit-and-assignment.in
rename to tests/functional/suites/repl/cases/inherit-and-assignment.in
diff --git a/tests/functional/repl/inherit-current-scope.expected b/tests/functional/suites/repl/cases/inherit-current-scope.expected
similarity index 100%
rename from tests/functional/repl/inherit-current-scope.expected
rename to tests/functional/suites/repl/cases/inherit-current-scope.expected
diff --git a/tests/functional/repl/inherit-current-scope.in b/tests/functional/suites/repl/cases/inherit-current-scope.in
similarity index 100%
rename from tests/functional/repl/inherit-current-scope.in
rename to tests/functional/suites/repl/cases/inherit-current-scope.in
diff --git a/tests/functional/repl/inherit-missing-shows-pos.expected b/tests/functional/suites/repl/cases/inherit-missing-shows-pos.expected
similarity index 100%
rename from tests/functional/repl/inherit-missing-shows-pos.expected
rename to tests/functional/suites/repl/cases/inherit-missing-shows-pos.expected
diff --git a/tests/functional/repl/inherit-missing-shows-pos.in b/tests/functional/suites/repl/cases/inherit-missing-shows-pos.in
similarity index 100%
rename from tests/functional/repl/inherit-missing-shows-pos.in
rename to tests/functional/suites/repl/cases/inherit-missing-shows-pos.in
diff --git a/tests/functional/repl/inherit-multiple-attrs.expected b/tests/functional/suites/repl/cases/inherit-multiple-attrs.expected
similarity index 100%
rename from tests/functional/repl/inherit-multiple-attrs.expected
rename to tests/functional/suites/repl/cases/inherit-multiple-attrs.expected
diff --git a/tests/functional/repl/inherit-multiple-attrs.in b/tests/functional/suites/repl/cases/inherit-multiple-attrs.in
similarity index 100%
rename from tests/functional/repl/inherit-multiple-attrs.in
rename to tests/functional/suites/repl/cases/inherit-multiple-attrs.in
diff --git a/tests/functional/repl/inherit-with-semicolon.expected b/tests/functional/suites/repl/cases/inherit-with-semicolon.expected
similarity index 100%
rename from tests/functional/repl/inherit-with-semicolon.expected
rename to tests/functional/suites/repl/cases/inherit-with-semicolon.expected
diff --git a/tests/functional/repl/inherit-with-semicolon.in b/tests/functional/suites/repl/cases/inherit-with-semicolon.in
similarity index 100%
rename from tests/functional/repl/inherit-with-semicolon.in
rename to tests/functional/suites/repl/cases/inherit-with-semicolon.in
diff --git a/tests/functional/repl/inherit.expected b/tests/functional/suites/repl/cases/inherit.expected
similarity index 100%
rename from tests/functional/repl/inherit.expected
rename to tests/functional/suites/repl/cases/inherit.expected
diff --git a/tests/functional/repl/inherit.in b/tests/functional/suites/repl/cases/inherit.in
similarity index 100%
rename from tests/functional/repl/inherit.in
rename to tests/functional/suites/repl/cases/inherit.in
diff --git a/tests/functional/repl/list-loaded-nothing-loaded.expected b/tests/functional/suites/repl/cases/list-loaded-nothing-loaded.expected
similarity index 100%
rename from tests/functional/repl/list-loaded-nothing-loaded.expected
rename to tests/functional/suites/repl/cases/list-loaded-nothing-loaded.expected
diff --git a/tests/functional/repl/list-loaded-nothing-loaded.in b/tests/functional/suites/repl/cases/list-loaded-nothing-loaded.in
similarity index 100%
rename from tests/functional/repl/list-loaded-nothing-loaded.in
rename to tests/functional/suites/repl/cases/list-loaded-nothing-loaded.in
diff --git a/tests/functional/repl/load-and-reload.expected b/tests/functional/suites/repl/cases/load-and-reload.expected
similarity index 75%
rename from tests/functional/repl/load-and-reload.expected
rename to tests/functional/suites/repl/cases/load-and-reload.expected
index 7f953e28613a..5d7f8f83648d 100644
--- a/tests/functional/repl/load-and-reload.expected
+++ b/tests/functional/suites/repl/cases/load-and-reload.expected
@@ -6,12 +6,12 @@ Added 3 variables.
 fromB, fromBFails, funcFromB
 
 nix-repl> :l ./does-not-exist.nix
-error: path '/path/to/tests/functional/repl/does-not-exist.nix' does not exist
+error: path '/path/to/repl/suite/cases/does-not-exist.nix' does not exist
 
 nix-repl> :p fromBFails
 error:
        … while calling the 'throw' builtin
-         at /path/to/tests/functional/repl/file-b.nix:2:16:
+         at /path/to/repl/suite/cases/file-b.nix:2:16:
             1| {
             2|   fromBFails = throw "b";
              |                ^
@@ -21,7 +21,7 @@ error:
 
 nix-repl> :doc funcFromB
 Function `funcFromB`\
-  … defined at /path/to/tests/functional/repl/file-b.nix:7:15
+  … defined at /path/to/repl/suite/cases/file-b.nix:7:15
 
 
 Some documentation.
diff --git a/tests/functional/repl/load-and-reload.in b/tests/functional/suites/repl/cases/load-and-reload.in
similarity index 100%
rename from tests/functional/repl/load-and-reload.in
rename to tests/functional/suites/repl/cases/load-and-reload.in
diff --git a/tests/functional/repl/missing-attr-shows-def.expected b/tests/functional/suites/repl/cases/missing-attr-shows-def.expected
similarity index 100%
rename from tests/functional/repl/missing-attr-shows-def.expected
rename to tests/functional/suites/repl/cases/missing-attr-shows-def.expected
diff --git a/tests/functional/repl/missing-attr-shows-def.in b/tests/functional/suites/repl/cases/missing-attr-shows-def.in
similarity index 100%
rename from tests/functional/repl/missing-attr-shows-def.in
rename to tests/functional/suites/repl/cases/missing-attr-shows-def.in
diff --git a/tests/functional/repl/multiline-string-binding.expected b/tests/functional/suites/repl/cases/multiline-string-binding.expected
similarity index 100%
rename from tests/functional/repl/multiline-string-binding.expected
rename to tests/functional/suites/repl/cases/multiline-string-binding.expected
diff --git a/tests/functional/repl/multiline-string-binding.in b/tests/functional/suites/repl/cases/multiline-string-binding.in
similarity index 100%
rename from tests/functional/repl/multiline-string-binding.in
rename to tests/functional/suites/repl/cases/multiline-string-binding.in
diff --git a/tests/functional/repl/multiline-string-function-binding.expected b/tests/functional/suites/repl/cases/multiline-string-function-binding.expected
similarity index 100%
rename from tests/functional/repl/multiline-string-function-binding.expected
rename to tests/functional/suites/repl/cases/multiline-string-function-binding.expected
diff --git a/tests/functional/repl/multiline-string-function-binding.in b/tests/functional/suites/repl/cases/multiline-string-function-binding.in
similarity index 100%
rename from tests/functional/repl/multiline-string-function-binding.in
rename to tests/functional/suites/repl/cases/multiline-string-function-binding.in
diff --git a/tests/functional/repl/multiple-bindings-same-line.expected b/tests/functional/suites/repl/cases/multiple-bindings-same-line.expected
similarity index 100%
rename from tests/functional/repl/multiple-bindings-same-line.expected
rename to tests/functional/suites/repl/cases/multiple-bindings-same-line.expected
diff --git a/tests/functional/repl/multiple-bindings-same-line.in b/tests/functional/suites/repl/cases/multiple-bindings-same-line.in
similarity index 100%
rename from tests/functional/repl/multiple-bindings-same-line.in
rename to tests/functional/suites/repl/cases/multiple-bindings-same-line.in
diff --git a/tests/functional/repl/nested-attr-path.expected b/tests/functional/suites/repl/cases/nested-attr-path.expected
similarity index 100%
rename from tests/functional/repl/nested-attr-path.expected
rename to tests/functional/suites/repl/cases/nested-attr-path.expected
diff --git a/tests/functional/repl/nested-attr-path.in b/tests/functional/suites/repl/cases/nested-attr-path.in
similarity index 100%
rename from tests/functional/repl/nested-attr-path.in
rename to tests/functional/suites/repl/cases/nested-attr-path.in
diff --git a/tests/functional/repl/pretty-print-idempotent.expected b/tests/functional/suites/repl/cases/pretty-print-idempotent.expected
similarity index 100%
rename from tests/functional/repl/pretty-print-idempotent.expected
rename to tests/functional/suites/repl/cases/pretty-print-idempotent.expected
diff --git a/tests/functional/repl/pretty-print-idempotent.in b/tests/functional/suites/repl/cases/pretty-print-idempotent.in
similarity index 100%
rename from tests/functional/repl/pretty-print-idempotent.in
rename to tests/functional/suites/repl/cases/pretty-print-idempotent.in
diff --git a/tests/functional/repl/pretty-print-idempotent.nix b/tests/functional/suites/repl/cases/pretty-print-idempotent.nix
similarity index 100%
rename from tests/functional/repl/pretty-print-idempotent.nix
rename to tests/functional/suites/repl/cases/pretty-print-idempotent.nix
diff --git a/tests/functional/repl/printing.expected b/tests/functional/suites/repl/cases/printing.expected
similarity index 100%
rename from tests/functional/repl/printing.expected
rename to tests/functional/suites/repl/cases/printing.expected
diff --git a/tests/functional/repl/printing.in b/tests/functional/suites/repl/cases/printing.in
similarity index 100%
rename from tests/functional/repl/printing.in
rename to tests/functional/suites/repl/cases/printing.in
diff --git a/tests/functional/suites/repl/characterisation.sh b/tests/functional/suites/repl/characterisation.sh
new file mode 100644
index 000000000000..a3f8ee5849b0
--- /dev/null
+++ b/tests/functional/suites/repl/characterisation.sh
@@ -0,0 +1,92 @@
+# shellcheck shell=bash
+# shellcheck source=/dev/null
+# shellcheck disable=SC2034
+
+{ set +x; } 2>/dev/null
+
+source "$_NIX_TEST_SOURCE_DIR"/common/characterisation/framework.sh
+
+testNumber=0
+
+# TODO: Convert all tests using characterisation tests to TAP and move this
+# to the framework.
+
+function skipTest() {
+    # TAP tests should always exit with 0.
+    exit 0
+}
+
+function reportDiffAcceptFailure() {
+    (( testNumber+=1 ))
+    local -r testName=$2
+    if [[ "${_NIX_TEST_ACCEPT-}" == 1 ]]; then
+        echo "ok $testNumber $testName # SKIP"
+    else
+        echo "not ok $testNumber $testName"
+    fi
+    echo "$1" >&2
+}
+
+function reportDiffAcceptSuccess() {
+    (( testNumber+=1 ))
+    echo "ok $testNumber $testName"
+}
+
+testDir=$PWD
+
+# TODO: move init to characterisation/framework.sh
+badDiff=0
+badExitCode=0
+
+testCases=( "$testDir"/cases/*.in )
+
+echo "TAP version 14"
+echo "1..${#testCases[@]}"
+
+for test in "${testCases[@]}"; do
+    test="$(basename "$test" .in)"
+    in="$testDir/cases/$test.in"
+    actual="$TEST_ROOT/$test.actual"
+    expected="$testDir/cases/$test.expected"
+    declare -a flags=()
+    if test -e "$testDir/cases/$test.flags"; then
+      read -r -a flags < "$testDir/cases/$test.flags"
+    fi
+
+    # Allow putting comments (lines starting with `# COM:`) in the test for
+    # documentation purposes. Regular comments are not skipped, since those are
+    # also interpreted by the repl.
+    inputWithoutComments=$(grep -Ev '^[[:space:]]*#[[:space:]]*COM:' "$in")
+
+    if [ -f "$testDir/cases/$test.in.debugexpr.nix" ]; then
+        if [[ "$test" =~ debugger-fail-.* ]]; then
+            expectedFail=1
+        elif [[ "$test" =~ debugger-okay-.* ]]; then
+            expectedFail=0
+        else
+            die "unexpected debugger test name: '$test', should be either debugger-okay-* or debugger-fail-*"
+        fi
+
+        set +e
+        (cd "$testDir/cases"; set +x; echo "$inputWithoutComments" | runDebugRepl "$testDir/cases/$test.in.debugexpr.nix" "$testDir" "${flags[@]}" 2>&1) > "$actual"
+        debuggerTestCode=$?
+        set -e
+
+        if ((expectedFail == 0 && debuggerTestCode != 0)); then
+            echo "test failed: $test (exit code $debuggerTestCode)" >&2
+            badExitCode=1
+        elif ((expectedFail == 1 && debuggerTestCode == 0)); then
+            echo "test failed: $test unexpectedly succeeded" >&2
+            badExitCode=1
+        fi
+    else
+        (cd "$testDir/cases"; set +x; echo "$inputWithoutComments" | runRepl "$testDir" "${flags[@]}" 2>&1) > "$actual" || {
+            echo "test failed: $test (exit code $?)" >&2
+            badExitCode=1
+        }
+    fi
+
+    diffAndAcceptInner "$test" "$actual" "$expected"
+done
+
+characterisationTestExit
diff --git a/tests/functional/suites/repl/common.sh b/tests/functional/suites/repl/common.sh
new file mode 100644
index 000000000000..a54361e6a093
--- /dev/null
+++ b/tests/functional/suites/repl/common.sh
@@ -0,0 +1,8 @@
+# shellcheck shell=bash
+
+{ set +x; } 2>/dev/null
+
+source "$_NIX_TEST_SOURCE_DIR"/common.sh
+source common/functions.sh
+
+set -x
diff --git a/tests/functional/suites/repl/common/functions.sh b/tests/functional/suites/repl/common/functions.sh
new file mode 100644
index 000000000000..f615788ddc80
--- /dev/null
+++ b/tests/functional/suites/repl/common/functions.sh
@@ -0,0 +1,99 @@
+# shellcheck shell=bash
+
+# Remove ANSI escape sequences. They can prevent grep from finding a match.
+stripColors () {
+    sed -E 's/\x1B\[([0-9]{1,2}(;[0-9]{1,2})?)?[m|K]//g'
+}
+
+testReplResponseGeneral () {
+    local grepMode commands expectedResponse response
+    grepMode="$1"; shift
+    commands="$1"; shift
+    # Expected response can contain newlines.
+    # grep can't handle multiline patterns, so replace newlines with TEST_NEWLINE
+    # in both expectedResponse and response.
+    # awk ORS always adds a trailing record separator, so we strip it with sed.
+    expectedResponse="$(printf '%s' "$1" | awk 1 ORS=TEST_NEWLINE | sed 's/TEST_NEWLINE$//')"; shift
+    # We don't need to strip trailing record separator here, since extra data is ok.
+    response="$(nix repl "$@" <<< "$commands" 2>&1 | stripColors | awk 1 ORS=TEST_NEWLINE)"
+    printf '%s' "$response" | grepQuiet "$grepMode" -s "$expectedResponse" \
+      || fail "$(echo "repl command set:
+
+$commands
+
+does not respond with:
+
+---
+$expectedResponse
+---
+
+but with:
+
+---
+$response
+---
+
+" | sed 's/TEST_NEWLINE/\n/g')"
+}
+
+testReplResponse () {
+    testReplResponseGeneral --basic-regexp "$@"
+}
+
+testReplResponseNoRegex () {
+    testReplResponseGeneral --fixed-strings "$@"
+}
+
+nixVersion="$(nix eval --impure --raw --expr 'builtins.nixVersion' --extra-experimental-features nix-command)"
+
+# TODO: write a repl interacter for testing. Papering over the differences between readline / editline and between platforms is a pain.
+
+# I couldn't get readline and editline to agree on the newline before the prompt,
+# so let's just force it to be one empty line.
+stripEmptyLinesBeforePrompt() {
+  # --null-data:  treat input as NUL-terminated instead of newline-terminated
+  sed --null-data 's/\n\n*nix-repl>/\n\nnix-repl>/g'
+}
+
+# We don't get a final prompt on darwin, so we strip this as well.
+stripFinalPrompt() {
+  # Strip the final prompt and/or any trailing spaces
+  sed --null-data \
+    -e 's/\(.*[^\n]\)\n\n*nix-repl>[ \n]*$/\1/' \
+    -e 's/[ \n]*$/\n/'
+}
+
+filterReplOutput () {
+  # That is right, we are also filtering out the testdir _without underscores_.
+  # This is crazy, but without it, GHA will fail to run the tests, showing paths
+  # _with_ underscores in the set -x log, but _without_ underscores in the
+  # supposed nix repl output. I have looked in a number of places, but I cannot
+  # find a mechanism that could cause this to happen.
+  local testDirNoUnderscores
+  local testDir="$1"
+
+  testDirNoUnderscores="${testDir//_/}"
+
+  stripColors \
+    | tr -d '\0' \
+    | stripEmptyLinesBeforePrompt \
+    | stripFinalPrompt \
+    | sed \
+      -e "s@$testDir@/path/to/repl/suite@g" \
+      -e "s@$testDirNoUnderscores@/path/to/repl/suite@g" \
+      -e "s@$nixVersion@<nix version>@g" \
+    | grep -vF $'warning: you don\'t have Internet access; disabling some network-dependent features' \
+    ;
+}
+
+runRepl () {
+    _NIX_TEST_RAW_MARKDOWN=1 \
+    _NIX_TEST_REPL_ECHO=1 \
+    nix repl "${@:2}" 2>&1 | filterReplOutput "$1"
+}
+
+runDebugRepl () {
+    _NIX_TEST_RAW_MARKDOWN=1 \
+    _NIX_TEST_REPL_ECHO=1 \
+    nix eval --file "$1" --debugger "${@:3}" 2>&1 | filterReplOutput "$2"
+}
diff --git a/tests/functional/suites/repl/edit-builtin-files.sh b/tests/functional/suites/repl/edit-builtin-files.sh
new file mode 100644
index 000000000000..3f441a8b701e
--- /dev/null
+++ b/tests/functional/suites/repl/edit-builtin-files.sh
@@ -0,0 +1,4 @@
+# shellcheck shell=bash
+
+EDITOR='cat' nix repl <<< ':e derivation' 2>&1 | grepQuiet 'derivationStrict'
+EDITOR='cat' nix repl <<< ':e <nix/fetchurl.nix>' 2>&1 | grepQuiet 'builtin:fetchurl'
diff --git a/tests/functional/suites/repl/function.nix b/tests/functional/suites/repl/function.nix
new file mode 100644
index 000000000000..3745ce9a6c70
--- /dev/null
+++ b/tests/functional/suites/repl/function.nix
@@ -0,0 +1,4 @@
+{
+  x ? 1,
+}:
+import ((builtins.getEnv "_NIX_TEST_SOURCE_DIR") + "/simple.nix")
diff --git a/tests/functional/suites/repl/meson.build b/tests/functional/suites/repl/meson.build
new file mode 100644
index 000000000000..8c782b361ed1
--- /dev/null
+++ b/tests/functional/suites/repl/meson.build
@@ -0,0 +1,15 @@
+suites += {
+  'name' : 'repl',
+  'deps' : [],
+  'tests' : [
+    'characterisation.sh',
+    'edit-builtin-files.sh',
+    'simple.sh',
+    'reload-flakes.sh',
+  ],
+  'workdir' : meson.current_source_dir(),
+  'env' : {'BASH_ENV' : meson.current_source_dir() / 'common.sh'},
+  'overrides' : {
+    'characterisation.sh' : {'protocol' : 'tap'},
+  },
+}
diff --git a/tests/functional/suites/repl/reload-flakes.sh b/tests/functional/suites/repl/reload-flakes.sh
new file mode 100644
index 000000000000..4361bd3b25f4
--- /dev/null
+++ b/tests/functional/suites/repl/reload-flakes.sh
@@ -0,0 +1,97 @@
+# shellcheck shell=bash
+
+# shellcheck disable=SC2164
+cd "$TEST_ROOT"
+
+# Doesn't do much without https://github.com/NixOS/nix/issues/10069
+enableFeatures 'flakes'
+
+mkdir -p flake && cat <<EOF > flake/flake.nix
+{
+    outputs = { self }: {
+        foo = 1;
+        bar.baz = 2;
+
+        changingThing = "beforeChange";
+    };
+}
+EOF
+
+testReplResponse '
+foo + baz
+' "3" ./flake ./flake\#bar
+
+# Test the `:reload` mechanism with flakes:
+# - Eval `./flake#changingThing`
+# - Modify the flake
+# - Re-eval it
+# - Check that the result has changed
+
+mkfifo repl_fifo
+touch repl_output
+nix repl ./flake < repl_fifo >> repl_output 2>&1 &
+repl_pid=$!
+exec 3>repl_fifo # Open fifo for writing
+echo "changingThing" >&3
+for i in $(seq 1 1000); do
+    if grep -q "beforeChange" repl_output; then
+        break
+    fi
+    cat repl_output
+    sleep 0.1
+done
+if [[ "$i" -eq 100 ]]; then
+    echo "Timed out waiting for beforeChange"
+    exit 1
+fi
+
+sed -i 's/beforeChange/afterChange/' flake/flake.nix
+
+# Send reload and second command
+echo ":reload" >&3
+echo "changingThing" >&3
+echo "exit" >&3
+exec 3>&- # Close fifo
+wait $repl_pid # Wait for process to finish
+grep -q "afterChange" repl_output
+
+# Regression: `:reload` on a flake loaded from a *git* work tree must pick up
+# uncommitted changes. Guards against the per-process workdir-info cache
+# pinning the tree to the rev seen on first load.
+if [[ $(type -p git) ]]; then
+    createGitRepo gitflake
+    cat > gitflake/flake.nix <<EOF
+{ outputs = { self }: { changingThing = "beforeChange"; }; }
+EOF
+    git -C gitflake add flake.nix
+    git -C gitflake commit -m init
+
+    rm -f repl_fifo repl_output
+    mkfifo repl_fifo
+    touch repl_output
+    nix repl ./gitflake < repl_fifo >> repl_output 2>&1 &
+    repl_pid=$!
+    exec 3>repl_fifo
+    echo "changingThing" >&3
+    for _ in $(seq 1 1000); do
+        grep -q "beforeChange" repl_output && break
+        sleep 0.1
+    done
+    grep -q "beforeChange" repl_output || fail "git flake didn't load"
+    sed -i 's/beforeChange/afterChange/' gitflake/flake.nix
+    echo ":reload" >&3
+    echo "changingThing" >&3
+    echo "exit" >&3
+    exec 3>&-
+    wait $repl_pid
+    grep -q "afterChange" repl_output || fail ":reload didn't pick up git work tree change"
+fi
+
+# Regression: a failed `:lf` must not be remembered for `:reload`,
+# and an error in one loaded file must not drop later ones from the reload list.
+testReplResponseNoRegex '
+:lf ./does-not-exist-flake
+:lf ./flake
+:r
+foo
+' '1'
diff --git a/tests/functional/suites/repl/simple.sh b/tests/functional/suites/repl/simple.sh
new file mode 100644
index 000000000000..df53fcf9771e
--- /dev/null
+++ b/tests/functional/suites/repl/simple.sh
@@ -0,0 +1,27 @@
+# shellcheck shell=bash
+
+testReplResponse '
+drvPath
+' '".*-simple.drv"' \
+    --file "$_NIX_TEST_SOURCE_DIR/simple.nix"
+
+testReplResponse '
+drvPath
+' '".*-simple.drv"' \
+    --file "$_NIX_TEST_SOURCE_DIR/simple.nix" --experimental-features 'ca-derivations'
+
+# `--file` autocalls, `--expr` does not.
+testReplResponse '
+drvPath
+' '".*-simple.drv"' \
+    --file "./function.nix"
+
+testReplResponse '
+' 'while evaluating an attribute set to be merged in the global scope' \
+    --expr '{ x ? 1 }: import '"$_NIX_TEST_SOURCE_DIR"'/simple.nix'
+
+# Don't prompt for more input when getting unexpected EOF in imported files.
+testReplResponse "
+import $_NIX_TEST_SOURCE_DIR/lang/parse-fail-eof-pos.nix
+" \
+    '.*error: syntax error, unexpected end of file.*'