Put features in sequence order, in the interest of greater clarity

README.md

@@ -205,10 +205,11 @@ resolution is obtained than by using a smaller _maxzoom_ or _detail_.
  * `-pk` or `--no-tile-size-limit`: Don't limit tiles to 500K bytes
  * `-pC` or `--no-tile-compression`: Don't compress the PBF vector tile data.
 
-### Temporary storage
+### System parameters
 
  * `-t` _directory_ or `--temporary-directory=`_directory_: Put the temporary files in _directory_.
    If you don't specify, it will use `/tmp`.
+ * `-j` _cpus_ or `--cpus=`_cpus_: Use the specified number of CPUs for parallel processing.
 
 ### Progress indicator
 
@@ -219,7 +220,7 @@ Environment
 -----------
 
 Tippecanoe ordinarily uses as many parallel threads as the operating system claims that CPUs are available.
-You can override this number by setting the `TIPPECANOE_MAX_THREADS` environmental variable.
+You can override this number by setting the `TIPPECANOE_MAX_THREADS` environmental variable or using the `--cpus` option.
 
 Example
 -------

geojson.cpp

@@ -493,7 +493,7 @@ int serialize_geometry(json_object *geometry, json_object *properties, json_obje
 	sf.feature_minzoom = 0;  // Will be filled in during index merging
 	sf.extent = (long long) extent;
 
-	if (prevent[P_INPUT_ORDER]) {
+	if (prevent[P_INPUT_ORDER] || additional[A_TAG_SEQUENCE]) {
 		sf.seq = *layer_seq;
 	} else {
 		sf.seq = 0;

main.cpp

@@ -118,10 +118,12 @@ void checkdisk(struct reader *r, int nreader) {
 	}
 };
 
-void init_cpus() {
+void init_cpus(size_t cpus) {
 	const char *TIPPECANOE_MAX_THREADS = getenv("TIPPECANOE_MAX_THREADS");
 
-	if (TIPPECANOE_MAX_THREADS != NULL) {
+	if (cpus > 0) {
+		CPUS = cpus;
+	} else if (TIPPECANOE_MAX_THREADS != NULL) {
 		CPUS = atoi(TIPPECANOE_MAX_THREADS);
 	} else {
 		CPUS = sysconf(_SC_NPROCESSORS_ONLN);
@@ -1204,7 +1206,7 @@ int read_input(std::vector<source> &sources, char *fname, int &maxzoom, int minz
 
 		int read_parallel_this = read_parallel ? '\n' : 0;
 
-		if (1) {
+		if (!prevent[P_MEMORY_MAP]) {
 			if (fstat(fd, &st) == 0) {
 				off = lseek(fd, 0, SEEK_CUR);
 				if (off >= 0) {
@@ -1280,7 +1282,7 @@ int read_input(std::vector<source> &sources, char *fname, int &maxzoom, int minz
 				unlink(readname);
 
 				volatile int is_parsing = 0;
-				long long ahead = 0;
+				size_t ahead = 0;
 				long long initial_offset = overall_offset;
 				pthread_t parallel_parser;
 				bool parser_created = false;
@@ -1289,19 +1291,28 @@ int read_input(std::vector<source> &sources, char *fname, int &maxzoom, int minz
 #define PARSE_MIN 10000000
 #define PARSE_MAX (1LL * 1024 * 1024 * 1024)
 
-				char buf[READ_BUF];
+				size_t parse_min = PARSE_MIN;
+				size_t read_buf = READ_BUF;
+				size_t parse_max = PARSE_MAX;
+				if (prevent[P_MEMORY_MAP]) {
+					parse_min = 2000;
+					read_buf = 5;
+					parse_max = 5;
+				}
+
+				char buf[read_buf];
 				int n;
 
-				while ((n = fread(buf, sizeof(char), READ_BUF, fp)) > 0) {
+				while ((n = fread(buf, sizeof(char), read_buf, fp)) > 0) {
 					fwrite_check(buf, sizeof(char), n, readfp, reading.c_str());
 					ahead += n;
 
-					if (buf[n - 1] == read_parallel_this && ahead > PARSE_MIN) {
+					if (buf[n - 1] == read_parallel_this && ahead > parse_min) {
 						// Don't let the streaming reader get too far ahead of the parsers.
 						// If the buffered input gets huge, even if the parsers are still running,
 						// wait for the parser thread instead of continuing to stream input.
 
-						if (is_parsing == 0 || ahead >= PARSE_MAX) {
+						if (is_parsing == 0 || ahead >= parse_max) {
 							if (parser_created) {
 								if (pthread_join(parallel_parser, NULL) != 0) {
 									perror("pthread_join 1088");
@@ -1988,8 +1999,6 @@ int main(int argc, char **argv) {
 	mtrace();
 #endif
 
-	init_cpus();
-
 	extern int optind;
 	extern char *optarg;
 	int i;
@@ -2013,6 +2022,7 @@ int main(int argc, char **argv) {
 	const char *attribution = NULL;
 	std::vector<source> sources;
 	bool guess_maxzoom = false;
+	size_t cpus = 0;
 
 	std::set<std::string> exclude, include;
 	std::map<std::string, int> attribute_types;
@@ -2111,8 +2121,9 @@ int main(int argc, char **argv) {
 		{"no-tile-size-limit", no_argument, &prevent[P_KILOBYTE_LIMIT], 1},
 		{"no-tile-compression", no_argument, &prevent[P_TILE_COMPRESSION], 1},
 
-		{"Temporary storage", 0, 0, 0},
+		{"System parameters", 0, 0, 0},
 		{"temporary-directory", required_argument, 0, 't'},
+		{"cpus", required_argument, 0, 'j'},
 
 		{"Progress indicator", 0, 0, 0},
 		{"quiet", no_argument, 0, 'q'},
@@ -2124,6 +2135,8 @@ int main(int argc, char **argv) {
 		{"check-polygons", no_argument, &additional[A_DEBUG_POLYGON], 1},
 		{"no-polygon-splitting", no_argument, &prevent[P_POLYGON_SPLIT], 1},
 		{"prefer-radix-sort", no_argument, &additional[A_PREFER_RADIX_SORT], 1},
+		{"tag-sequence", no_argument, &additional[A_TAG_SEQUENCE], 1},
+		{"no-memory-mapping", no_argument, &prevent[P_MEMORY_MAP], 1},
 
 		{0, 0, 0, 0},
 	};
@@ -2307,6 +2320,10 @@ int main(int argc, char **argv) {
 			}
 			break;
 
+		case 'j':
+			cpus = atoi(optarg);
+			break;
+
 		case 'g':
 			gamma = atof(optarg);
 			break;
@@ -2401,6 +2418,8 @@ int main(int argc, char **argv) {
 		}
 	}
 
+	init_cpus(cpus);
+
 	files_open_at_start = open("/dev/null", O_RDONLY);
 	close(files_open_at_start);
 

man/tippecanoe.1

@@ -250,11 +250,13 @@ the line or polygon within one tile unit of its proper location. You can probabl
 .IP \(bu 2
 \fB\fC\-pC\fR or \fB\fC\-\-no\-tile\-compression\fR: Don't compress the PBF vector tile data.
 .RE
-.SS Temporary storage
+.SS System parameters
 .RS
 .IP \(bu 2
 \fB\fC\-t\fR \fIdirectory\fP or \fB\fC\-\-temporary\-directory=\fR\fIdirectory\fP: Put the temporary files in \fIdirectory\fP\&.
 If you don't specify, it will use \fB\fC/tmp\fR\&.
+.IP \(bu 2
+\fB\fC\-j\fR \fIcpus\fP or \fB\fC\-\-cpus=\fR\fIcpus\fP: Use the specified number of CPUs for parallel processing.
 .RE
 .SS Progress indicator
 .RS
@@ -266,7 +268,7 @@ If you don't specify, it will use \fB\fC/tmp\fR\&.
 .SH Environment
 .PP
 Tippecanoe ordinarily uses as many parallel threads as the operating system claims that CPUs are available.
-You can override this number by setting the \fB\fCTIPPECANOE_MAX_THREADS\fR environmental variable.
+You can override this number by setting the \fB\fCTIPPECANOE_MAX_THREADS\fR environmental variable or using the \fB\fC\-\-cpus\fR option.
 .SH Example
 .PP
 .RS

options.hpp

@@ -14,6 +14,7 @@
 #define A_DROP_SMALLEST_AS_NEEDED ((int) 'n')
 #define A_GRID_LOW_ZOOMS ((int) 'L')
 #define A_DETECT_WRAPAROUND ((int) 'w')
+#define A_TAG_SEQUENCE ((int) 'S')
 
 #define P_SIMPLIFY ((int) 's')
 #define P_SIMPLIFY_LOW ((int) 'S')
@@ -26,6 +27,7 @@
 #define P_DUPLICATION ((int) 'D')
 #define P_TINY_POLYGON_REDUCTION ((int) 't')
 #define P_TILE_COMPRESSION ((int) 'C')
+#define P_MEMORY_MAP ((int) 'M')
 
 extern int prevent[256];
 extern int additional[256];

tile.cpp

@@ -1730,6 +1730,13 @@ long long write_tile(FILE *geoms, long long *geompos_in, char *metabase, char *s
 
 				decode_meta(layer_features[x].m, layer_features[x].keys, layer_features[x].values, layer_features[x].stringpool, layer, feature);
 
+				if (additional[A_TAG_SEQUENCE]) {
+					mvt_value v;
+					v.type = mvt_sint;
+					v.numeric_value.sint_value = layer_features[x].original_seq;
+					layer.tag(feature, "tippecanoe_sequence", v);
+				}
+
 				if (additional[A_CALCULATE_FEATURE_DENSITY]) {
 					int glow = 255;
 					if (layer_features[x].spacing > 0) {