diff --git a/app/include/lwip/app/espconn_tcp.h b/app/include/lwip/app/espconn_tcp.h
index 76aa4b8c1a..3da9dabde0 100644
--- a/app/include/lwip/app/espconn_tcp.h
+++ b/app/include/lwip/app/espconn_tcp.h
@@ -16,6 +16,9 @@
 #define   espconn_manual_recv_disabled(espconn)  (((espconn)->pcommon.espconn_opt & ESPCONN_MANUALRECV) != 0)
 #define   espconn_manual_recv_enabled(espconn)  (((espconn)->pcommon.espconn_opt & ESPCONN_MANUALRECV) == 0)
 
+extern int ets_task();
+extern int ets_post();
+
 /******************************************************************************
  * FunctionName : espconn_kill_oldest_pcb
  * Description  : A oldest incoming connection has been killed.
diff --git a/app/include/lwip/mem.h b/app/include/lwip/mem.h
index 825d222611..52362f3acc 100644
--- a/app/include/lwip/mem.h
+++ b/app/include/lwip/mem.h
@@ -45,6 +45,14 @@ extern "C" {
 
 typedef size_t mem_size_t;
 
+void *pvPortMalloc (size_t sz, const char *, unsigned, bool);
+void vPortFree (void *p, const char *, unsigned);
+void *pvPortZalloc (size_t sz, const char *, unsigned);
+void *pvPortRealloc (void *p, size_t n, const char *, unsigned);
+void* pvPortCalloc(size_t count,size_t size,const char *,unsigned);
+void* pvPortCallocIram(size_t count,size_t size,const char *,unsigned);
+void *pvPortZallocIram (size_t sz, const char *, unsigned);
+
 /* aliases for C library malloc() */
 #define mem_init()
 /* in case C library malloc() needs extra protection,
diff --git a/app/include/mbedtls/platform.h b/app/include/mbedtls/platform.h
index 1f907d6fed..3a164897e2 100644
--- a/app/include/mbedtls/platform.h
+++ b/app/include/mbedtls/platform.h
@@ -58,6 +58,9 @@ extern "C" {
  * \{
  */
 
+void *pvPortCalloc(unsigned int count, unsigned int size, const char*, unsigned);
+void vPortFree (void *p, const char *, unsigned);
+
 #if !defined(MBEDTLS_PLATFORM_NO_STD_FUNCTIONS)
 #include <stdio.h>
 #include <stdlib.h>
diff --git a/app/include/sys/socket.h b/app/include/sys/socket.h
index a57378dc3e..937cacc042 100644
--- a/app/include/sys/socket.h
+++ b/app/include/sys/socket.h
@@ -63,6 +63,9 @@ typedef enum{
 	NETCONN_STATE_SUMNUM
 }netconn_state;
 
+extern int __attribute__((weak)) espconn_mbedtls_parse_internal(int socket, sint8 error);
+extern int __attribute__((weak)) espconn_mbedtls_parse_thread(int socket, int event, int error);
+
 #if (!defined(lwIP_unlikely))
 #define lwIP_unlikely(Expression)	!!(Expression)
 #endif
@@ -330,4 +333,8 @@ uint32_t lwip_getul(char *str);
 #define close(s)              lwip_close(s)
 #define getul(s)			  lwip_getul(s)
 
+extern int system_overclock(void);
+extern int system_restoreclock(void);
+extern char *sys_itoa(int n);
+
 #endif /* ESPCONN_SOCKT_H_ */
diff --git a/app/include/user_config.h b/app/include/user_config.h
index 928d58847f..3540d35d25 100644
--- a/app/include/user_config.h
+++ b/app/include/user_config.h
@@ -42,21 +42,34 @@
 //#define DISABLE_STARTUP_BANNER
 
 
-// Three separate build variants are now supported. The main difference is in the
-// processing of numeric data types.  If LUA_NUMBER_INTEGRAL is defined, then
+// When using Lua 5.1, two different builds are now supported.
+// The main difference is in the // processing of numeric data types.
+// If LUA_NUMBER_INTEGRAL is defined, then
 // all numeric calculations are done in integer, with divide being an integer
-// operations, and decimal fraction constants are illegal.  Otherwise all
-// numeric operations use floating point, though they are exact for integer
-// expressions < 2^53.
-
-// The main advantage of INTEGRAL builds is that the basic internal storage unit,
-// the TValue, is 8 bytes long.  We have now reduced the size of FP TValues to
-// 12 bytes rather than the previous 16 as this gives a material RAM saving with
-// no performance loss.  However, you can define LUA_DWORD_ALIGNED_TVALUES and
-// this will force 16 byte TValues on FP builds.
+// operation, and decimal fraction constants are illegal.
+// Otherwise all floating point operations use doubles. All integer values
+// can be represented exactly in floating point.
 
 //#define LUA_NUMBER_INTEGRAL
-//#define LUA_DWORD_ALIGNED_TVALUES
+
+// When using Lua 5.3, two different builds are now supported. 
+// The main difference is in the processing of numeric data types.
+// If LUA_NUMBER_64BITS is defined, then doubles are used to hold floating
+// point numbers. Integers under 2^53 are representable exactly in doubles.
+// Integers are held in 64-bit variables. 
+// Otherwise all floating point operations use floats. Only integers under 2^24
+// can be represented exactly in floating point. Integers are represented in 32 bit variables.
+// Note that Lua 5.3 also supports Integers natively, but you have to be careful 
+// not to promote an integer to a floating point variable if you are using a float build
+// as you can lose precision.
+
+//#define LUA_NUMBER_64BITS
+
+// The main advantage of INTEGRAL builds and non 64BITS builds is that the basic internal
+// storage unit, the TValue, is 8 bytes long.  For 64BITS builds, we have now reduced
+// the size of FP TValues to 12 bytes rather than the previous 16 as this gives a
+// material RAM saving with no performance loss.
+//
 
 
 // The Lua Flash Store (LFS) allows you to store Lua code in Flash memory and
@@ -138,7 +151,7 @@
 
 // The net module optionally offers net info functionnality. Uncomment the following
 // to enable the functionnality.
-#define NET_PING_ENABLE 
+#define NET_PING_ENABLE
 
 // The WiFi module optionally offers an enhanced level of WiFi connection
 // management, using internal timer callbacks.  Whilst many Lua developers
@@ -251,7 +264,7 @@
 #define READLINE_INTERVAL        80
 #define STRBUF_DEFAULT_INCREMENT  3
 #define LUA_USE_BUILTIN_DEBUG_MINIMAL // for debug.getregistry() and debug.traceback()
- 
+
 #if defined(DEVELOPMENT_TOOLS) && defined(DEVELOPMENT_USE_GDB)
 extern void LUA_DEBUG_HOOK (void);
 #define lua_assert(x)    ((x) ? (void) 0 : LUA_DEBUG_HOOK ())
diff --git a/app/lua/luaconf.h b/app/lua/luaconf.h
index 95c6117e1f..13e1e3008d 100644
--- a/app/lua/luaconf.h
+++ b/app/lua/luaconf.h
@@ -181,6 +181,10 @@
   #endif // #if !defined LUA_INTEGRAL_LONGLONG
 #endif // #if !defined LUA_NUMBER_INTEGRAL
 
+#ifdef LUA_NUMBER_64BITS
+#error Lua 5.1 does not support 64 bit inetegers.
+#endif
+
 /*
 @@ LUA_API is a mark for all core API functions.
 @@ LUALIB_API is a mark for all standard library functions.
diff --git a/app/lua53/ldump.c b/app/lua53/ldump.c
index 9fca72b8ad..135389bc63 100644
--- a/app/lua53/ldump.c
+++ b/app/lua53/ldump.c
@@ -101,7 +101,7 @@ static void DumpNumber (lua_Number x, DumpState *D) {
 ** 0TTTNNNN or 1TTTNNNN (1NNNNNNN)* 0NNNNNNN
 */
 static void DumpIntTT (lu_byte tt, lua_Integer y, DumpState *D) {
-  int x = y < 0 ? -(y + 1) : y;
+  lua_Integer x = y < 0 ? -(y + 1) : y;
   lu_byte buf[sizeof(lua_Integer) + 3];
   lu_byte *b = buf + sizeof(buf) - 1;
   *b-- = x & 0x7f;  x >>= 7;
diff --git a/app/lua53/lobject.h b/app/lua53/lobject.h
index c7d0c630ff..6d3815c2f6 100644
--- a/app/lua53/lobject.h
+++ b/app/lua53/lobject.h
@@ -134,12 +134,17 @@ typedef union Value {
 
 #define TValuefields	Value value_; int tt_
 
+#ifdef LUA_USE_ESP
+#  pragma pack(4)
+#endif
 
 typedef struct lua_TValue {
   TValuefields;
 } TValue;
 
-
+#ifdef LUA_USE_ESP
+#  pragma pack()
+#endif
 
 /* macro defining a nil value */
 #define NILCONSTANT	{NULL}, LUA_TNIL
diff --git a/app/lua53/luaconf.h b/app/lua53/luaconf.h
index 28c5041566..fa717fd825 100644
--- a/app/lua53/luaconf.h
+++ b/app/lua53/luaconf.h
@@ -85,9 +85,17 @@
 #  define LUA_FLOAT_TYPE	LUA_FLOAT_DOUBLE
 #endif
 #endif
-#  define LUA_INT_TYPE    LUA_INT_INT
+#ifdef LUA_NUMBER_64BITS
+#  define LUA_FLOAT_TYPE  LUA_FLOAT_DOUBLE
+#  define LUA_INT_TYPE    LUA_INT_LONGLONG
+#else
 #  define LUA_FLOAT_TYPE  LUA_FLOAT_FLOAT
-//#  define LUA_FLOAT_TYPE  LUA_FLOAT_DOUBLE
+#  define LUA_INT_TYPE    LUA_INT_INT
+#endif
+
+#ifdef LUA_NUMBER_INTEGRAL
+#error LUA_NUMBER_INTEGRAL is not supported in LUA5.3 builds
+#endif
 
 /*
 ** Configuration for Paths.
diff --git a/app/mbedtls/app/Makefile b/app/mbedtls/app/Makefile
index fa5dc92fd6..55c0baa5ae 100644
--- a/app/mbedtls/app/Makefile
+++ b/app/mbedtls/app/Makefile
@@ -17,6 +17,7 @@ GEN_LIBS = libapp.a
 
 endif
 
+STD_CFLAGS=-std=gnu11
 
 #############################################################
 # Configuration i.e. compile options etc.
diff --git a/app/mbedtls/app/espconn_mbedtls.c b/app/mbedtls/app/espconn_mbedtls.c
index 82beaec9e6..15282b7bb4 100644
--- a/app/mbedtls/app/espconn_mbedtls.c
+++ b/app/mbedtls/app/espconn_mbedtls.c
@@ -42,6 +42,8 @@ static const char mem_debug_file[] ICACHE_RODATA_ATTR = __FILE__;
 
 #include "sys/socket.h"
 #include "sys/espconn_mbedtls.h"
+#include "lwip/app/espconn_tcp.h"
+
 
 static os_event_t lwIPThreadQueue[lwIPThreadQueueLen];
 static bool lwIPThreadFlag = false;
diff --git a/app/mbedtls/app/lwIPSocket.c b/app/mbedtls/app/lwIPSocket.c
index 8c798f1c8f..4b35141398 100644
--- a/app/mbedtls/app/lwIPSocket.c
+++ b/app/mbedtls/app/lwIPSocket.c
@@ -34,6 +34,7 @@
 
 #include "ets_sys.h"
 #include "os_type.h"
+#include <ctype.h>
 
 #include "lwip/mem.h"
 #include "sys/socket.h"
@@ -42,6 +43,9 @@
 static const char mem_debug_file[] ICACHE_RODATA_ATTR = __FILE__;
 #endif
 
+void *pvPortZalloc (size_t sz, const char *, unsigned);
+void vPortFree (void *p, const char *, unsigned);
+
 /** The global array of available sockets */
 static lwIP_sock sockets[NUM_SOCKETS];
 
diff --git a/app/net/nodemcu_mdns.c b/app/net/nodemcu_mdns.c
index 5750a9b73a..119d7a38f7 100644
--- a/app/net/nodemcu_mdns.c
+++ b/app/net/nodemcu_mdns.c
@@ -1001,7 +1001,7 @@ mdns_set_servicename(const char *name) {
 	char tmpBuf[128];
 	os_sprintf(tmpBuf, "_%s._tcp.local", name);
 	if (service_name_with_suffix) {
-	  os_free(service_name_with_suffix);
+	  os_free((void *) service_name_with_suffix);
 	}
 	service_name_with_suffix = strdup(tmpBuf);
 }
diff --git a/docs/build.md b/docs/build.md
index e9f04e5f97..18c9489242 100644
--- a/docs/build.md
+++ b/docs/build.md
@@ -76,13 +76,31 @@ editing `BIT_RATE_DEFAULT` in `app/include/user_config.h`:
 Note that, by default, the firmware runs an auto-baudrate detection algorithm so that typing a few characters at boot time will cause
 the firmware to lock onto that baud rate (between 1200 and 230400).
 
-### Integer build
-By default a build will be generated supporting floating-point variables.
+### Double build (Lua 5.3 only)
+By default a build will be generated supporting floating point variables (floats) and integers.
+To increase the precision of the floating point variables, a double build can be created. This
+is also the default in the Lua 5.1 builds. The downside is that more memory is consumed when
+storing variables.
+You can change this
+either by uncommenting `LUA_NUMBER_64BITS` in `app/include/user_config.h`:
+
+```c
+//#define LUA_NUMBER_64BITS
+```
+
+OR by overriding this with the `make` command 
+
+```
+make EXTRA_CCFLAGS="-DLUA_NUMBER_64BITS ....
+```
+
+### Integer build (Lua 5.1 only)
+By default a build will be generated supporting floating-point variables (doubles).
 To reduce memory size an integer build can be created.  You can change this
 either by uncommenting `LUA_NUMBER_INTEGRAL` in `app/include/user_config.h`:
 
 ```c
-#define LUA_NUMBER_INTEGRAL
+//#define LUA_NUMBER_INTEGRAL
 ```
 
 OR by overriding this with the `make` command as it's [done during the CI
diff --git a/docs/compiling.md b/docs/compiling.md
index 731ee547bb..700a44f747 100644
--- a/docs/compiling.md
+++ b/docs/compiling.md
@@ -57,11 +57,30 @@ facilitates simple OTA updates to an LFS based Lua application; the absolute for
 facilitates factory installation of LFS based applications.
 
 Also note that the `app/lua/luac_cross` make and Makefile can be executed to build
-just the `luac.cross` image.  You must first ensure that the following option in
-`app/include/user_config.h` is matched to your target configuration:
+just the `luac.cross` image.  You must first ensure that the following options in
+`app/include/user_config.h` are matched to your target configuration:
 
 ```c
-//#define LUA_NUMBER_INTEGRAL       // uncomment if you want an integer build
+// When using Lua 5.1, two different builds are now supported.
+// The main difference is in the // processing of numeric data types.
+// If LUA_NUMBER_INTEGRAL is defined, then
+// all numeric calculations are done in integer, with divide being an integer
+// operation, and decimal fraction constants are illegal.
+// Otherwise all floating point operations use doubles. All integer values
+// can be represented exactly in floating point.
+
+//#define LUA_NUMBER_INTEGRAL
+
+// When using Lua 5.3, two different builds are now supported. 
+// If LUA_NUMBER_64BITS is defined, then doubles are used to hold floating
+// point numbers and integers are 64 bits long. Integers smaller than 2^53 can be i
+// converted to doubles and back without any loss of precision.
+// Otherwise all floating point operations use floats, and integers are 32-bit.
+// Only integers under 2^24 can be represented exactly in floating point.
+// Note that Lua 5.3 also supports Integers natively, but you have to be careful 
+// not to promote an integer to a floating point variable as you can lose precision.
+
+//#define LUA_NUMBER_64BITS
 ```
 
 Note that the use of LFS and the LFS region size is now configured through the partition table.
diff --git a/tools/update_buildinfo.sh b/tools/update_buildinfo.sh
index 7caa09e5d5..747caee92a 100755
--- a/tools/update_buildinfo.sh
+++ b/tools/update_buildinfo.sh
@@ -38,8 +38,12 @@ cat > $TEMPFILE << EndOfMessage
 #ifdef LUA_NUMBER_INTEGRAL
 #define BUILDINFO_BUILD_TYPE "integer"
 #else
+#ifdef LUA_NUMBER_64BITS
+#define BUILDINFO_BUILD_TYPE "double"
+#else
 #define BUILDINFO_BUILD_TYPE "float"
 #endif
+#endif
 
 #define USER_PROLOG "$USER_PROLOG"
 #define BUILDINFO_BRANCH "$BRANCH"