Restructure benchmark layout (prepare to add new benchmarks).

Update pre-generated benchmark files (trivial line directive changes).

Disable regeneration of Kleenex files, as Kleenex is not actively maintained and
fails to build with the recent cabal and Haskell ecosystem.

.github/workflows/benchmarks.yml

@@ -88,20 +88,20 @@ jobs:
         run: cmake --build .build
 
       - name: Run submatch DFA AOT benchmark
-        working-directory: .build/benchmarks/submatch_dfa_aot
+        working-directory: .build/benchmarks/c/submatch
         run: |
           ./run.py --quickverify \
             && ./run.py --repetitions 5 --output results.json \
             && cat results.json
 
       - name: Run submatch DFA JIT benchmark
-        working-directory: .build/benchmarks/submatch_dfa_jit
-        run: ./bench_submatch_dfa_jit
+        working-directory: .build/benchmarks/c/submatch_libre2c/jit
+        run: ./bench_submatch_jit
 
       - name: Run submatch Java benchmark (ReTdfa)
-        working-directory: .build/benchmarks/submatch_java
+        working-directory: .build/benchmarks/c/submatch_libre2c/java
         run: ./run.py
 
       - name: Run submatch NFA benchmark
-        working-directory: .build/benchmarks/submatch_nfa
+        working-directory: .build/benchmarks/c/submatch_libre2c/nfa
         run: ./bench_submatch_nfa

.gitmodules

@@ -1,3 +1,3 @@
-[submodule "benchmarks/submatch_java/REgen"]
-	path = benchmarks/submatch_java/REgen
+[submodule "benchmarks/c/submatch_libre2c/java/REgen"]
+	path = benchmarks/c/submatch_libre2c/java/REgen
 	url = https://github.com/skvadrik/REgen.git

CMakeLists.txt

@@ -633,9 +633,9 @@ if (RE2C_BUILD_LIBS)
     endif()
 
     if(RE2C_BUILD_BENCHMARKS)
-        add_subdirectory(benchmarks/submatch_nfa)
-        add_subdirectory(benchmarks/submatch_dfa_jit)
-        add_subdirectory(benchmarks/submatch_java)
+        add_subdirectory(benchmarks/c/submatch_libre2c/nfa)
+        add_subdirectory(benchmarks/c/submatch_libre2c/jit)
+        add_subdirectory(benchmarks/c/submatch_libre2c/java)
     endif()
 else()
     # empty check target
@@ -645,7 +645,7 @@ else()
 endif()
 
 if(RE2C_BUILD_BENCHMARKS)
-    add_subdirectory(benchmarks/submatch_dfa_aot)
+    add_subdirectory(benchmarks/c/submatch)
 endif()
 
 if(RE2C_BUILD_TESTS)

Makefile.am

@@ -761,8 +761,8 @@ TESTS = \
 
 # benchmarks
 if WITH_BENCHMARKS
-SUBDIRS += benchmarks/submatch_dfa_aot
-EXTRA_DIST += benchmarks/submatch_dfa_aot
+SUBDIRS += benchmarks/c/submatch
+EXTRA_DIST += benchmarks/c
 endif
 
 # libre2c

Makefile.lib.am

@@ -169,12 +169,12 @@ check_PROGRAMS += test_libre2c
 
 # benchmarks
 if WITH_BENCHMARKS
-SUBDIRS += benchmarks/submatch_nfa
-SUBDIRS += benchmarks/submatch_dfa_jit
-EXTRA_DIST += benchmarks/submatch_nfa
-EXTRA_DIST += benchmarks/submatch_dfa_jit
+SUBDIRS += benchmarks/c/submatch_libre2c/nfa
+SUBDIRS += benchmarks/c/submatch_libre2c/jit
+EXTRA_DIST += benchmarks/c/submatch_libre2c/nfa
+EXTRA_DIST += benchmarks/c/submatch_libre2c/jit
 if WITH_JAVA
-SUBDIRS += benchmarks/submatch_java
-EXTRA_DIST += benchmarks/submatch_java
+SUBDIRS += benchmarks/c/submatch_libre2c/java
+EXTRA_DIST += benchmarks/c/submatch_libre2c/java
 endif # WITH_JAVA
 endif # WITH_BENCHMARKS

benchmarks/common/common.cc → benchmarks/c/common/common.cc

@@ -1,6 +1,6 @@
 #include <string.h>
 
-#include "benchmarks/common/common.h"
+#include "benchmarks/c/common/common.h"
 #include "lib/regex.h"
 
 #ifdef HAVE_RE2_RE2_H

benchmarks/submatch_dfa_aot/CMakeLists.txt → benchmarks/c/submatch/CMakeLists.txt

@@ -53,7 +53,7 @@ file(MAKE_DIRECTORY "${GEN_DIR}/kleenex/")
 set(KLEENEX "${ENG_DIR}/kleenex/kexc")
 add_custom_command(
     OUTPUT "${KLEENEX}"
-    COMMAND "${CMAKE_CURRENT_SOURCE_DIR}/engines/kleenex/getkleenex.sh"
+    COMMAND "${CMAKE_CURRENT_SOURCE_DIR}/../engines/kleenex/getkleenex.sh"
     WORKING_DIRECTORY "${ENG_DIR}/kleenex/"
 )
 # Masked benchmarks for which Kleenex either generates very large output
@@ -71,23 +71,24 @@ foreach(bench ${KLEENEX_BENCHMARKS})
     set(src_file "${SRC_DIR}/kleenex/${bench}.kex")
     set(gen_file "${GEN_DIR}/kleenex/${bench}.c")
     set(pregen_file "${PREGEN_DIR}/kleenex/${bench}.c")
-    # optionally regenerate kleenex benchmarks if RE2C_REGEN_BENCHMARKS is set
-    if(RE2C_REGEN_BENCHMARKS)
-        file(RELATIVE_PATH rel_src_file "${CMAKE_CURRENT_BINARY_DIR}" "${src_file}")
-        file(RELATIVE_PATH rel_gen_file "${CMAKE_CURRENT_BINARY_DIR}" "${gen_file}")
-        add_custom_command(
-            OUTPUT "${gen_file}"
-            COMMAND "${KLEENEX}" compile "${rel_src_file}" --srcout "${rel_gen_file}" --act=false --la=true
-            COMMAND "${CMAKE_COMMAND}" -E copy_if_different "${gen_file}" "${pregen_file}"
-            DEPENDS "${src_file}" "${KLEENEX}"
-        )
-    else()
+    # don't regenerate kleenex benchmarks, as kleenex is not actively developed
+    # and it cannot be built with the recent cabal / haskell ecosystem
+    # if(RE2C_REGEN_BENCHMARKS)
+    #     file(RELATIVE_PATH rel_src_file "${CMAKE_CURRENT_BINARY_DIR}" "${src_file}")
+    #     file(RELATIVE_PATH rel_gen_file "${CMAKE_CURRENT_BINARY_DIR}" "${gen_file}")
+    #     add_custom_command(
+    #         OUTPUT "${gen_file}"
+    #         COMMAND "${KLEENEX}" compile "${rel_src_file}" --srcout "${rel_gen_file}" --act=false --la=true
+    #         COMMAND "${CMAKE_COMMAND}" -E copy_if_different "${gen_file}" "${pregen_file}"
+    #         DEPENDS "${src_file}" "${KLEENEX}"
+    #     )
+    # else()
         add_custom_command(
             OUTPUT "${gen_file}"
             COMMAND "${CMAKE_COMMAND}" -E copy "${pregen_file}" "${gen_file}"
             DEPENDS "${pregen_file}"
         )
-    endif()
+    # endif()
     list(APPEND GEN "${gen_file}")
 endforeach()
 
@@ -97,7 +98,7 @@ file(MAKE_DIRECTORY "${GEN_DIR}/ragel/")
 set(RAGEL "${ENG_DIR}/ragel/ragel7")
 add_custom_command(
     OUTPUT "${RAGEL}"
-    COMMAND "${CMAKE_CURRENT_SOURCE_DIR}/engines/ragel/getragel7.sh"
+    COMMAND "${CMAKE_CURRENT_SOURCE_DIR}/../engines/ragel/getragel7.sh"
     WORKING_DIRECTORY "${ENG_DIR}/ragel/"
 )
 foreach(bench ${BENCHMARKS})
@@ -131,7 +132,7 @@ set(RE2C "${CMAKE_BINARY_DIR}/re2c")
 set(RE2C3 "${ENG_DIR}/re2c/re2c3")
 add_custom_command(
     OUTPUT "${RE2C3}"
-    COMMAND "${CMAKE_CURRENT_SOURCE_DIR}/engines/re2c/getre2c3.sh"
+    COMMAND "${CMAKE_CURRENT_SOURCE_DIR}/../engines/re2c/getre2c3.sh"
     WORKING_DIRECTORY "${ENG_DIR}/re2c/"
 )
 set(RE2C_FLAGS "--reusable" "--tags" "--no-generation-date" "--no-version")

benchmarks/submatch_dfa_aot/Makefile.am → benchmarks/c/submatch/Makefile.am

@@ -128,6 +128,7 @@ $(GEN_RE2C_EOF_TDFA1): $(GEN_DIR)/%-eof-tdfa1.c: $(SRC_DIR)/%.re $(COMMON_RE2C)
 	$(AM_V_GEN)$(RE2C) $(RE2C_FLAGS_EOF) $< -o $@
 	$(AM_V_at)if ! cmp -s $@ $(PREGEN_DIR)/re2c/$(@F) ; then cp -f $@ $(PREGEN_DIR)/re2c/$(@F) ; fi
 
+
 # optionally regenerate ragel, kleenex and re2c-3.0 benchmarks
 # (TDFA(0) and staDFA algorithms were removed in re2c versions after 3.0)
 
@@ -158,10 +159,15 @@ $(GEN_RAGEL): $(GEN_DIR)/%.c: $(SRC_DIR)/%.rl $(COMMON_RAGEL) $(RAGEL)
 	$(AM_V_GEN)$(RAGEL) -G2 $< -o $@
 	$(AM_V_at)if ! cmp -s $@ $(PREGEN_DIR)/ragel/$(@F) ; then cp -f $@ $(PREGEN_DIR)/ragel/$(@F) ; fi
 
-$(GEN_KLEENEX): $(GEN_DIR)/%.c: $(SRC_DIR)/%.kex $(KLEENEX)
+# don't regenerate kleenex benchmarks, as kleenex is not actively developed
+# and it cannot be built with the recent cabal / haskell ecosystem
+# $(GEN_KLEENEX): $(GEN_DIR)/%.c: $(SRC_DIR)/%.kex $(KLEENEX)
+# 	$(AM_V_at)mkdir -p $(@D)
+# 	$(AM_V_GEN)$(KLEENEX) compile $< --srcout $@ --act=false --la=true
+# 	$(AM_V_at)if ! cmp -s $@ $(PREGEN_DIR)/kleenex/$(@F) ; then cp -f $@ $(PREGEN_DIR)/kleenex/$(@F) ; fi
+$(GEN_KLEENEX): $(GEN_DIR)/%: $(PREGEN_DIR)/%
 	$(AM_V_at)mkdir -p $(@D)
-	$(AM_V_GEN)$(KLEENEX) compile $< --srcout $@ --act=false --la=true
-	$(AM_V_at)if ! cmp -s $@ $(PREGEN_DIR)/kleenex/$(@F) ; then cp -f $@ $(PREGEN_DIR)/kleenex/$(@F) ; fi
+	$(AM_V_GEN)cp -f $< $@
 
 else
 
@@ -186,15 +192,15 @@ $(DAT): $(DAT_DIR)/%/big: $(DAT_DIR)/%/small
 	$(AM_V_GEN)( cd $(@D) && ../../$(srcdir)/data/gen.py )
 
 $(KLEENEX):
-	$(AM_V_at)mkdir -p $(@D) && cp $(srcdir)/engines/kleenex/getkleenex.sh $(@D)
+	$(AM_V_at)mkdir -p $(@D) && cp $(srcdir)/../engines/kleenex/getkleenex.sh $(@D)
 	$(AM_V_GEN)( cd $(@D) && ./getkleenex.sh )
 
 $(RAGEL):
-	$(AM_V_at)mkdir -p $(@D) && cp $(srcdir)/engines/ragel/getragel7.sh $(@D)
+	$(AM_V_at)mkdir -p $(@D) && cp $(srcdir)/../engines/ragel/getragel7.sh $(@D)
 	$(AM_V_GEN)( cd $(@D) && ./getragel7.sh )
 
 $(RE2C3):
-	$(AM_V_at)mkdir -p $(@D) && cp $(srcdir)/engines/re2c/getre2c3.sh $(@D)
+	$(AM_V_at)mkdir -p $(@D) && cp $(srcdir)/../engines/re2c/getre2c3.sh $(@D)
 	$(AM_V_GEN)( cd $(@D) && ./getre2c3.sh )
 
 clean-local: