use disjoint sets for struct and lang-el identifiers

Using disjoint sets allows us to verify at runtime that we are not mixing trees
and structs. Some of this is still left over from when we used trees for
streams and lists and things. Need to build up some internal consistency
checks.

8 files changed, 26 insertions, 15 deletions

diff --git a/src/compiler.h b/src/compiler.h
index cf23bc9c..085fdd1d 100644
--- a/src/compiler.h
+++ b/src/compiler.h
@@ -907,6 +907,7 @@ struct Compiler
 
 	int nextLelId;
 	int firstNonTermId;
+	int structInbuiltId;
 
 	LangEl **langElIndex;
 	PdaState *actionDestState;
diff --git a/src/input.c b/src/input.c
index 5ae2a78d..cb5522c3 100644
--- a/src/input.c
+++ b/src/input.c
@@ -1113,7 +1113,7 @@ stream_t *colm_stream_new_struct( program_t *prg )
 	struct colm_stream *stream = (struct colm_stream*) malloc( memsize );
 	memset( stream, 0, memsize );
 	colm_struct_add( prg, (struct colm_struct *)stream );
-	stream->id = STRUCT_INBUILT_ID;
+	stream->id = prg->rtd->struct_inbuilt_id;
 	stream->destructor = &colm_stream_destroy;
 	return stream;
 }
diff --git a/src/list.c b/src/list.c
index ebcc4618..245e3331 100644
--- a/src/list.c
+++ b/src/list.c
@@ -205,7 +205,7 @@ list_t *colm_list_new( struct colm_program *prg )
 	struct colm_list *list = (struct colm_list*) malloc( memsize );
 	memset( list, 0, memsize );
 	colm_struct_add( prg, (struct colm_struct *)list );
-	list->id = STRUCT_INBUILT_ID;
+	list->id = prg->rtd->struct_inbuilt_id;
 	list->destructor = &colm_list_destroy;
 	return list;
 }
diff --git a/src/pdabuild.cc b/src/pdabuild.cc
index 6ab45471..2ce1cecb 100644
--- a/src/pdabuild.cc
+++ b/src/pdabuild.cc
@@ -242,9 +242,14 @@ void Compiler::makeLangElIds()
 
 void Compiler::makeStructElIds()
 {
-	int nextId = 0;
+	/* Start at the next lang el id and go up from there. Using disjoint sets
+	 * allows us to verify that a tree is a tree and struct is a struct because
+	 * the ID field is at the same offset. */
+	int nextId = nextLelId;
 	for ( StructElList::Iter sel = structEls; sel.lte(); sel++ )
 		sel->id = nextId++;
+
+	structInbuiltId = nextId;
 }
 
 void Compiler::refNameSpace( LangEl *lel, Namespace *nspace )
@@ -1709,6 +1714,7 @@ void Compiler::makeRuntimeData()
 	runtimeData->no_token_id = noTokenLangEl->id;
 	runtimeData->global_id = globalSel->id;
 	runtimeData->argv_el_id = argvElSel->id;
+	runtimeData->struct_inbuilt_id = structInbuiltId;
 
 	runtimeData->fsm_execute = &internalFsmExecute;
 	runtimeData->send_named_lang_el = &internalSendNamedLangEl;
diff --git a/src/pdacodegen.cc b/src/pdacodegen.cc
index ef1c1ced..2e08ee3e 100644
--- a/src/pdacodegen.cc
+++ b/src/pdacodegen.cc
@@ -510,6 +510,7 @@ void PdaCodeGen::writeRuntimeData( colm_sections *runtimeData, struct pda_tables
 		"	" << runtimeData->no_token_id << ",\n"
 		"	" << runtimeData->global_id << ",\n"
 		"	" << runtimeData->argv_el_id << ",\n"
+		"	" << runtimeData->struct_inbuilt_id << ",\n"
 		"	&fsm_execute,\n"
 		"	&sendNamedLangEl,\n"
 		"	&initBindings,\n"
diff --git a/src/program.h b/src/program.h
index 8138c393..e5cfbeea 100644
--- a/src/program.h
+++ b/src/program.h
@@ -98,6 +98,7 @@ struct colm_sections
 	long no_token_id;
 	long global_id;
 	long argv_el_id;
+	long struct_inbuilt_id;
 
 	void (*fsm_execute)( struct pda_run *pda_run, struct stream_impl *input_stream );
 	void (*send_named_lang_el)( struct colm_program *prg, tree_t **tree,
diff --git a/src/struct.c b/src/struct.c
index 11b3e2ec..0a6fd781 100644
--- a/src/struct.c
+++ b/src/struct.c
@@ -61,24 +61,28 @@ struct colm_struct *colm_struct_new_size( program_t *prg, int size )
 
 struct colm_struct *colm_struct_new( program_t *prg, int id )
 {
-	struct colm_struct *s = colm_struct_new_size( prg, prg->rtd->sel_info[id].size );
+	struct colm_struct *s = colm_struct_new_size( prg, prg->rtd->sel_info[id - prg->rtd->num_lang_els].size );
 	s->id = id;
 	return s;
 }
 
+struct struct_el_info *colm_sel_info( program_t *prg, int id )
+{
+	return &prg->rtd->sel_info[id - prg->rtd->num_lang_els];
+}
+
 void colm_struct_delete( program_t *prg, tree_t **sp, struct colm_struct *el )
 {
-	if ( el->id == STRUCT_INBUILT_ID ) {
+	if ( el->id == prg->rtd->struct_inbuilt_id ) {
 		colm_destructor_t destructor = ((struct colm_inbuilt*)el)->destructor;
 		if ( destructor != 0 )
 			(*destructor)( prg, sp, el );
 	}
-
-	if ( el->id >= 0 ) { 
-		short *t = prg->rtd->sel_info[el->id].trees;
-		int i, len = prg->rtd->sel_info[el->id].trees_len;
-		for ( i = 0; i < len; i++ ) {
-			tree_t *tree = colm_struct_get_field( el, tree_t*, t[i] );
+	else {
+		int tree_i;
+		struct struct_el_info *sel = colm_sel_info( prg, el->id );
+		for ( tree_i = 0; tree_i < sel->trees_len; tree_i++ ) {
+			tree_t *tree = colm_struct_get_field( el, tree_t*, sel->trees[tree_i] );
 			colm_tree_downref( prg, sp, tree );
 		}
 	}
@@ -110,7 +114,7 @@ parser_t *colm_parser_new( program_t *prg, struct generic_info *gi, int reducer
 	memset( parser, 0, memsize );
 	colm_struct_add( prg, (struct colm_struct*) parser );
 
-	parser->id = STRUCT_INBUILT_ID;
+	parser->id = prg->rtd->struct_inbuilt_id;
 	parser->destructor = &colm_parser_destroy;
 	parser->pda_run = pda_run;
 
@@ -136,7 +140,7 @@ map_t *colm_map_new( struct colm_program *prg )
 	struct colm_map *map = (struct colm_map*) malloc( memsize );
 	memset( map, 0, memsize );
 	colm_struct_add( prg, (struct colm_struct *)map );
-	map->id = STRUCT_INBUILT_ID;
+	map->id = prg->rtd->struct_inbuilt_id;
 	return map;
 }
 
diff --git a/src/struct.h b/src/struct.h
index ccc8b442..bafbfb64 100644
--- a/src/struct.h
+++ b/src/struct.h
@@ -173,8 +173,6 @@ struct colm_struct *colm_map_get( struct colm_program *prg, map_t *map,
 struct colm_struct *colm_construct_generic( struct colm_program *prg, long generic_id );
 struct colm_struct *colm_construct_reducer( struct colm_program *prg, long generic_id, int reducer_id );
 
-#define STRUCT_INBUILT_ID -1
-
 #if defined(__cplusplus)
 }
 #endif