Made sure we properly check the size.

btypes.h

@@ -0,0 +1,29 @@
+
+
+#ifndef BOWTIE_INDEX_TYPES_H
+#define	BOWTIE_INDEX_TYPES_H
+
+#ifdef BOWTIE_64BIT_INDEX
+#define OFF_MASK 0xffffffffffffffff
+#define OFF_LEN_MASK 0xc000000000000000
+#define LS_SIZE 0x100000000000000
+#define OFF_SIZE 8
+
+typedef uint64_t TIndexOffU;
+typedef int64_t TIndexOff;
+
+#else
+#define OFF_MASK 0xffffffff
+#define OFF_LEN_MASK 0xc0000000
+#define LS_SIZE 0x10000000
+#define OFF_SIZE 4
+
+typedef uint32_t TIndexOffU;
+typedef int TIndexOff;
+
+#endif /* BOWTIE_64BIT_INDEX */
+
+extern const std::string gEbwt_ext;
+
+#endif	/* BOWTIE_INDEX_TYPES_H */
+

ref_read.cpp

@@ -235,7 +235,7 @@ fastaRefReadSizes(vector<FileBuf*>& in,
                   BitpairOutFileBuf* bpout,
                   int& numSeqs)
 {
-	uint32_t unambigTot = 0;
+	TIndexOffU unambigTot = 0;
 	uint32_t bothTot = 0;
 	assert_gt(in.size(), 0);
 	uint32_t both = 0, unambig = 0;
@@ -258,9 +258,13 @@ fastaRefReadSizes(vector<FileBuf*>& in,
 			if(rec.len == 0) rec.first = false;
 #endif
 			if((unambigTot + rec.len) < unambigTot) {
+#ifdef BOWTIE_64BIT_INDEX
 				cerr << "Error: Reference sequence has more than 2^32-1 characters!  Please divide the" << endl
-				     << "reference into batches or chunks of about 3.6 billion characters or less each" << endl
-				     << "and index each independently." << endl;
+				     << "reference into smaller chunks and index each independently." << endl;
+#else
+				cerr << "Error: Reference sequence has more than 2^32-1 characters!  Please try to" << endl
+				     << "build a large index instead using the appropiate options." << endl;
+#endif
 				throw 1;
 			}
 			// Add the length of this record.

word_io.h

@@ -6,6 +6,7 @@
 #include <fstream>
 #include "assert_helpers.h"
 #include "endian_swap.h"
+#include "btypes.h"
 
 /**
  * Write a 32-bit unsigned to an output stream being careful to