exercism · keiravillekode · Aug 14, 2026 · Aug 13, 2026
@@ -41,8 +41,8 @@ RUN rm -rf .git build vm src misc Factor.app \
     GNUmakefile Nmakefile LICENSE.txt README.md \
     build.sh build.cmd unmaintained
 
-# Keep only the extra/ source that sequences.extras needs to compile on demand:
-# itself plus extra/assocs.extras and extra/shuffle.
+# Keep only the extra/ source needed to compile sequences.extras and
+# assocs.extras on demand: those two plus extra/shuffle.
 RUN find extra -mindepth 1 -maxdepth 1 \
     ! -name sequences ! -name assocs ! -name shuffle \
     -exec rm -rf {} + && \

@@ -47,7 +47,11 @@ tmp_dir=$(mktemp -d -t "factor-runner-${slug}-XXXXX")
 trap 'rm -rf "${tmp_dir}"' EXIT
 cp -r "${solution_dir}/." "${tmp_dir}"
 stripped_tests="${tmp_dir}/${slug}/${slug}-tests.factor"
-awk '!/^STOP-HERE$/' "${stripped_tests}" > "${stripped_tests}.new"
+# Blank each STOP-HERE line rather than deleting it: a blank line is a no-op
+# to Factor, and keeping the line count intact means the line numbers Factor
+# reports for failures still match the file the student submitted.
+awk '{ if ($0 ~ /^STOP-HERE$/) print ""; else print }' \
+    "${stripped_tests}" > "${stripped_tests}.new"
 mv "${stripped_tests}.new" "${stripped_tests}"
 
 # Run Factor; capture combined stdout/stderr.
@@ -131,7 +135,7 @@ src_tests=$(awk "${awk_json}"'
 
 # 2. Parse Factor stdout into NDJSON segments and failures:
 #    segments: {"type":"segment","idx":N,"failed":bool,"name":"...","output":"..."}
-#    failures: {"type":"failure","line_no":N,"message":"..."}
+#    failures: {"type":"failure","line_no":N,"location":"...","message":"..."}
 parsed=$(printf '%s\n' "${raw_output}" | awk "${awk_json}"'
     function close_segment(   out, i) {
         if (idx == 0) return
@@ -147,8 +151,8 @@ parsed=$(printf '%s\n' "${raw_output}" | awk "${awk_json}"'
         for (i = 1; i <= fail_n; i++) body = body (i > 1 ? "\n" : "") fail[i]
         sub(/^\n+/, "", body)
         sub(/\n+$/, "", body)
-        printf "{\"type\":\"failure\",\"line_no\":%d,\"message\":%s}\n",
-            fail_line, json_str(body)
+        printf "{\"type\":\"failure\",\"line_no\":%d,\"location\":%s,\"message\":%s}\n",
+            fail_line, json_str(fail_loc), json_str(body)
     }
     # Factor renders each test-word name into a title, e.g.:
     #   unit-test            → "Unit Test:"
@@ -221,8 +225,14 @@ parsed=$(printf '%s\n' "${raw_output}" | awk "${awk_json}"'
             n = substr($0, RSTART, RLENGTH)
             gsub(/[^0-9]/, "", n)
             fail_line = n + 0
+            # exercism-tools prints "<path>: <line#>"; collapse it to the
+            # conventional "<path>:<line#>" so the student can paste it
+            # into an editor. The path is already relative to the solution
+            # root, so it is safe to show as-is.
+            fail_loc = substr($0, 1, RSTART - 1) ":" fail_line
         } else {
             fail_line = 0
+            fail_loc = ""
         }
         fail_n = 0
         delete fail
@@ -276,15 +286,16 @@ jq -n \
     --argjson segs   "$(printf '%s\n' "${segments}"   | jq -s '.')" \
     --argjson fails  "$(printf '%s\n' "${failures}"   | jq -s '.')" \
     '
-    ($fails | map({(.line_no|tostring): .message}) | add // {}) as $fail_by_line
+    ($fails | map({(.line_no|tostring): .}) | add // {}) as $fail_by_line
     | $segs | sort_by(.idx)
     | to_entries
     | map(
         .value as $seg
         | (.key) as $i
         | ($srcs[$i] // null) as $src
         | ($src.line_no | tostring) as $ln
-        | ($fail_by_line[$ln] // null) as $msg
+        | ($fail_by_line[$ln] // null) as $fail
+        | ($fail.message // null) as $msg
         | (
             if $seg.failed then
                 if ($msg // "" | startswith("=== Expected:")) then "fail"
@@ -300,7 +311,11 @@ jq -n \
             test_code: ($src.test_code // ""),
           }
           + (if $src.task_id then {task_id: $src.task_id} else {} end)
-          + (if $seg.failed then {message: ($msg // "test failed")} else {} end)
+          + (if $seg.failed then
+               {message: (if $msg == null then "test failed"
+                          elif ($fail.location // "") == "" then $msg
+                          else $fail.location + "\n" + $msg end)}
+             else {} end)
           + (if $seg.output != "" then {output: ($seg.output[0:500])} else {} end)
       )
     | (if all(.status == "pass") then "pass" else "fail" end) as $top

@@ -1,6 +1,7 @@
 USING: accessors command-line continuations debugger io kernel
-       lexer namespaces sequences source-files.errors.debugger
-       system tools.test vocabs vocabs.loader ;
+       lexer namespaces prettyprint.config sequences
+       source-files.errors.debugger system tools.test vocabs
+       vocabs.loader ;
 IN: exercism-tools
 
 SYNTAX: STOP-HERE
@@ -9,26 +10,23 @@ SYNTAX: STOP-HERE
 SYNTAX: TASK:
     lexer get next-line ;
 
-! Label the test that follows with its description. The marker lets the
-! wrapper strip this line from captured output and attach it to the next
-! test as a name, rather than leaving it in the previous test's output.
+! Label the test that follows with its description.
 : description ( str -- )
     "###DESC### " write print ;
 
-! Print one failure block in a stable, parser-friendly form. Bracketed by
-! markers so a wrapper can split the stream reliably and avoid Factor's
-! noisy callstack output (which is interleaved with subsequent failures).
+! Print one failure block in a stable, parser-friendly form.
 :: print-failure ( failure -- )
     "###FAIL_BEGIN###" print
     failure error-location print
-    failure error>> [ error. ] [ 2drop ] recover
+    [ failure error>> [ error. ] [ 2drop ] recover ] without-limits
     "###FAIL_END###" print
     flush ;
 
 : print-failures ( -- )
     test-failures get [ print-failure ] each ;
 
 : run-exercism-tests ( -- )
+    vocab-roots [ "." prefix ] change-global
     command-line get first
     [ require ] [ test ] bi
     test-failures get empty?

@@ -6,13 +6,13 @@
       "name": "greet returns hello",
       "status": "fail",
       "test_code": "{ \"hello\" } [ greet ] unit-test",
-      "message": "=== Expected:\n\"hello\"\n=== Got:\n\"wrong\""
+      "message": "all-fail/all-fail-tests.factor:5\n=== Expected:\n\"hello\"\n=== Got:\n\"wrong\""
     },
     {
       "name": "greet returns world",
       "status": "fail",
       "test_code": "{ \"world\" } [ greet ] unit-test",
-      "message": "=== Expected:\n\"world\"\n=== Got:\n\"wrong\""
+      "message": "all-fail/all-fail-tests.factor:10\n=== Expected:\n\"world\"\n=== Got:\n\"wrong\""
     }
   ]
 }
@@ -1,6 +1,7 @@
 USING: accessors command-line continuations debugger io kernel
-       lexer namespaces sequences source-files.errors.debugger
-       system tools.test vocabs vocabs.loader ;
+       lexer namespaces prettyprint.config sequences
+       source-files.errors.debugger system tools.test vocabs
+       vocabs.loader ;
 IN: exercism-tools
 
 SYNTAX: STOP-HERE
@@ -9,26 +10,23 @@ SYNTAX: STOP-HERE
 SYNTAX: TASK:
     lexer get next-line ;
 
-! Label the test that follows with its description. The marker lets the
-! wrapper strip this line from captured output and attach it to the next
-! test as a name, rather than leaving it in the previous test's output.
+! Label the test that follows with its description.
 : description ( str -- )
     "###DESC### " write print ;
 
-! Print one failure block in a stable, parser-friendly form. Bracketed by
-! markers so a wrapper can split the stream reliably and avoid Factor's
-! noisy callstack output (which is interleaved with subsequent failures).
+! Print one failure block in a stable, parser-friendly form.
 :: print-failure ( failure -- )
     "###FAIL_BEGIN###" print
     failure error-location print
-    failure error>> [ error. ] [ 2drop ] recover
+    [ failure error>> [ error. ] [ 2drop ] recover ] without-limits
     "###FAIL_END###" print
     flush ;
 
 : print-failures ( -- )
     test-failures get [ print-failure ] each ;
 
 : run-exercism-tests ( -- )
+    vocab-roots [ "." prefix ] change-global
     command-line get first
     [ require ] [ test ] bi
     test-failures get empty?

@@ -1,6 +1,7 @@
 USING: accessors command-line continuations debugger io kernel
-       lexer namespaces sequences source-files.errors.debugger
-       system tools.test vocabs vocabs.loader ;
+       lexer namespaces prettyprint.config sequences
+       source-files.errors.debugger system tools.test vocabs
+       vocabs.loader ;
 IN: exercism-tools
 
 SYNTAX: STOP-HERE
@@ -9,26 +10,23 @@ SYNTAX: STOP-HERE
 SYNTAX: TASK:
     lexer get next-line ;
 
-! Label the test that follows with its description. The marker lets the
-! wrapper strip this line from captured output and attach it to the next
-! test as a name, rather than leaving it in the previous test's output.
+! Label the test that follows with its description.
 : description ( str -- )
     "###DESC### " write print ;
 
-! Print one failure block in a stable, parser-friendly form. Bracketed by
-! markers so a wrapper can split the stream reliably and avoid Factor's
-! noisy callstack output (which is interleaved with subsequent failures).
+! Print one failure block in a stable, parser-friendly form.
 :: print-failure ( failure -- )
     "###FAIL_BEGIN###" print
     failure error-location print
-    failure error>> [ error. ] [ 2drop ] recover
+    [ failure error>> [ error. ] [ 2drop ] recover ] without-limits
     "###FAIL_END###" print
     flush ;
 
 : print-failures ( -- )
     test-failures get [ print-failure ] each ;
 
 : run-exercism-tests ( -- )
+    vocab-roots [ "." prefix ] change-global
     command-line get first
     [ require ] [ test ] bi
     test-failures get empty?

@@ -73,7 +73,7 @@
       "status": "fail",
       "test_code": "{ 6 } [ 4 [ 6 roll-die ] play-seeded ] unit-test",
       "task_id": 5,
-      "message": "=== Expected:\n6\n=== Got:\n2"
+      "message": "concept-multiline/concept-multiline-tests.factor:62\n=== Expected:\n6\n=== Got:\n2"
     },
     {
       "name": "Test 13",

@@ -1,6 +1,7 @@
 USING: accessors command-line continuations debugger io kernel
-       lexer namespaces sequences source-files.errors.debugger
-       system tools.test vocabs vocabs.loader ;
+       lexer namespaces prettyprint.config sequences
+       source-files.errors.debugger system tools.test vocabs
+       vocabs.loader ;
 IN: exercism-tools
 
 SYNTAX: STOP-HERE
@@ -9,26 +10,23 @@ SYNTAX: STOP-HERE
 SYNTAX: TASK:
     lexer get next-line ;
 
-! Label the test that follows with its description. The marker lets the
-! wrapper strip this line from captured output and attach it to the next
-! test as a name, rather than leaving it in the previous test's output.
+! Label the test that follows with its description.
 : description ( str -- )
     "###DESC### " write print ;
 
-! Print one failure block in a stable, parser-friendly form. Bracketed by
-! markers so a wrapper can split the stream reliably and avoid Factor's
-! noisy callstack output (which is interleaved with subsequent failures).
+! Print one failure block in a stable, parser-friendly form.
 :: print-failure ( failure -- )
     "###FAIL_BEGIN###" print
     failure error-location print
-    failure error>> [ error. ] [ 2drop ] recover
+    [ failure error>> [ error. ] [ 2drop ] recover ] without-limits
     "###FAIL_END###" print
     flush ;
 
 : print-failures ( -- )
     test-failures get [ print-failure ] each ;
 
 : run-exercism-tests ( -- )
+    vocab-roots [ "." prefix ] change-global
     command-line get first
     [ require ] [ test ] bi
     test-failures get empty?

@@ -1,5 +1,5 @@
 {
   "version": 3,
   "status": "error",
-  "message": "/opt/test-runner/tests/concept-not-parsing/concept-not-parsing/concept-not-parsing.factor\n\n6: : cube ( n -- n^3 ) ?!?! ;\n                           ^\nNo word named '?!?!' found in current vocabulary search path"
+  "message": "concept-not-parsing/concept-not-parsing.factor\n\n6: : cube ( n -- n^3 ) ?!?! ;\n                           ^\nNo word named '?!?!' found in current vocabulary search path"
 }
@@ -1,6 +1,7 @@
 USING: accessors command-line continuations debugger io kernel
-       lexer namespaces sequences source-files.errors.debugger
-       system tools.test vocabs vocabs.loader ;
+       lexer namespaces prettyprint.config sequences
+       source-files.errors.debugger system tools.test vocabs
+       vocabs.loader ;
 IN: exercism-tools
 
 SYNTAX: STOP-HERE
@@ -9,26 +10,23 @@ SYNTAX: STOP-HERE
 SYNTAX: TASK:
     lexer get next-line ;
 
-! Label the test that follows with its description. The marker lets the
-! wrapper strip this line from captured output and attach it to the next
-! test as a name, rather than leaving it in the previous test's output.
+! Label the test that follows with its description.
 : description ( str -- )
     "###DESC### " write print ;
 
-! Print one failure block in a stable, parser-friendly form. Bracketed by
-! markers so a wrapper can split the stream reliably and avoid Factor's
-! noisy callstack output (which is interleaved with subsequent failures).
+! Print one failure block in a stable, parser-friendly form.
 :: print-failure ( failure -- )
     "###FAIL_BEGIN###" print
     failure error-location print
-    failure error>> [ error. ] [ 2drop ] recover
+    [ failure error>> [ error. ] [ 2drop ] recover ] without-limits
     "###FAIL_END###" print
     flush ;
 
 : print-failures ( -- )
     test-failures get [ print-failure ] each ;
 
 : run-exercism-tests ( -- )
+    vocab-roots [ "." prefix ] change-global
     command-line get first
     [ require ] [ test ] bi
     test-failures get empty?

@@ -21,50 +21,50 @@
       "status": "fail",
       "test_code": "{ 27 } [ \"cubing 3\" print 3 cube ] unit-test",
       "task_id": 2,
-      "message": "=== Expected:\n27\n=== Got:\n12",
+      "message": "concept-partial-fail/concept-partial-fail-tests.factor:12\n=== Expected:\n27\n=== Got:\n12",
       "output": "cubing 3"
     },
     {
       "name": "Test 4",
       "status": "fail",
       "test_code": "{ 64 } [ 4 cube ] unit-test",
       "task_id": 2,
-      "message": "=== Expected:\n64\n=== Got:\n20"
+      "message": "concept-partial-fail/concept-partial-fail-tests.factor:13\n=== Expected:\n64\n=== Got:\n20"
     },
     {
       "name": "Test 5",
       "status": "error",
       "test_code": "{ { } } [ { } but-first ] unit-test",
       "task_id": 3,
-      "message": "Cannot create slice: from > to\nfrom 1\nto   0\nseq  { }"
+      "message": "concept-partial-fail/concept-partial-fail-tests.factor:16\nCannot create slice: from > to\nfrom 1\nto   0\nseq  { }"
     },
     {
       "name": "Test 6",
       "status": "error",
       "test_code": "{ 0 } [ 4 0 / ] unit-test",
       "task_id": 4,
-      "message": "Division by zero\nx 4"
+      "message": "concept-partial-fail/concept-partial-fail-tests.factor:19\nDivision by zero\nx 4"
     },
     {
       "name": "Test 7",
       "status": "error",
       "test_code": "{ 0 } [ 10 { 1 2 3 } nth ] unit-test",
       "task_id": 5,
-      "message": "Sequence index out of bounds\nindex 10\nseq   { 1 2 3 }"
+      "message": "concept-partial-fail/concept-partial-fail-tests.factor:22\nSequence index out of bounds\nindex 10\nseq   { 1 2 3 }"
     },
     {
       "name": "Test 8",
       "status": "error",
       "test_code": "{ \"text\" } [ 5 label ] unit-test",
       "task_id": 6,
-      "message": "Generic word label does not define a method for the fixnum class.\nDispatching on object: 5"
+      "message": "concept-partial-fail/concept-partial-fail-tests.factor:25\nGeneric word label does not define a method for the fixnum class.\nDispatching on object: 5"
     },
     {
       "name": "Test 9",
       "status": "error",
       "test_code": "{ 0 } [ 5 [ drop ] call( x -- x ) ] unit-test",
       "task_id": 7,
-      "message": "Quotation's stack effect does not match call site\nquot      [ drop ]\ncall-site ( x -- x )"
+      "message": "concept-partial-fail/concept-partial-fail-tests.factor:28\nQuotation's stack effect does not match call site\nquot      [ drop ]\ncall-site ( x -- x )"
     }
   ]
 }