--- a/Makefile.src.in
+++ b/Makefile.src.in
@@ -20,6 +20,7 @@
 TTD          = !!TTD!!
 STRGEN       = !!STRGEN!!
 ENDIAN_CHECK = !!ENDIAN_CHECK!!
+DEPEND       = !!DEPEND!!
 ENDIAN_FORCE = !!ENDIAN_FORCE!!
 OS           = !!OS!!
 STAGE        = !!STAGE!!
@@ -167,11 +168,18 @@
 -include Makefile.dep
 endif
 
+ifeq ("$(SRC_OBJS_DIR)/$(DEPEND)","$(MAKEDEPEND)")
+DEP := $(MAKEDEPEND)
+$(SRC_OBJS_DIR)/$(DEPEND): $(SRC_DIR)/depend/depend.cpp
+	$(E) '$(STAGE) Compiling and linking $(DEPEND)'
+	$(Q)$(CXX_HOST) -o $@ $<
+endif
+
 # Make sure that only 'make depend' ALWAYS triggers a recheck
 ifeq ($(filter depend, $(MAKECMDGOALS)),)
-Makefile.dep: $(FILE_DEP) $(SRCS:%=$(SRC_DIR)/%) $(CONFIG_CACHE_SOURCE)
+Makefile.dep: $(FILE_DEP) $(SRCS:%=$(SRC_DIR)/%) $(CONFIG_CACHE_SOURCE) $(DEP)
 else
-Makefile.dep: FORCE
+Makefile.dep: $(DEP) FORCE
 endif
 	$(E) '$(STAGE) DEP CHECK (all files)'
 	$(Q)rm -f Makefile.dep.tmp
@@ -291,7 +299,7 @@
 
 clean:
 	$(E) '$(STAGE) Cleaning up object files'
-	$(Q)rm -f $(DEPS) $(OBJS) $(TTD) $(TTD:%=$(BIN_DIR)/%) $(CONFIG_CACHE_COMPILER) $(CONFIG_CACHE_LINKER) $(CONFIG_CACHE_ENDIAN) $(CONFIG_CACHE_SOURCE) $(ENDIAN_TARGETS)
+	$(Q)rm -f $(DEPS) $(OBJS) $(TTD) $(DEPEND) $(TTD:%=$(BIN_DIR)/%) $(CONFIG_CACHE_COMPILER) $(CONFIG_CACHE_LINKER) $(CONFIG_CACHE_ENDIAN) $(CONFIG_CACHE_SOURCE) $(ENDIAN_TARGETS)
 
 mrproper: clean
 	$(Q)rm -f $(SRC_DIR)/rev.cpp $(SRC_DIR)/ottdres.rc

--- a/config.lib
+++ b/config.lib
@@ -61,7 +61,8 @@
 	with_cocoa="1"
 	with_zlib="1"
 	with_png="1"
-	with_makedepend="1"
+	enable_builtin_depend="1"
+	with_makedepend="0"
 	with_direct_music="1"
 	with_sort="1"
 	with_iconv="1"
@@ -127,6 +128,7 @@
 		with_cocoa
 		with_zlib
 		with_png
+		enable_builtin_depend
 		with_makedepend
 		with_direct_music
 		with_sort
@@ -339,6 +341,10 @@
 			--without-psp-config)         with_psp_config="0";;
 			--with-psp-config=*)          with_psp_config="$optarg";;
 
+			--disable-builtin-depend)     enable_builtin_depend="0";;
+			--enable-builtin-depend)      enable_builtin_depend="2";;
+			--enable-builtin-depend=*)    enable_builtin_depend="$optarg";;
+
 			--with-makedepend)            with_makedepend="2";;
 			--without-makedepend)         with_makedepend="0";;
 			--with-makedepend=*)          with_makedepend="$optarg";;
@@ -511,6 +517,14 @@
 		log 1 "checking strip... disabled"
 	fi
 	check_lipo
+
+	if [ "$enable_builtin_depend" != "0" ]; then
+		log 1 "checking builtin depend... yes"
+		makedepend="\$(SRC_OBJS_DIR)/\$(DEPEND)"
+	else
+		log 1 "checking builtin depend... no"
+	fi
+
 	check_makedepend
 	detect_cputype
 
@@ -1418,7 +1432,7 @@
 	# Btw, this almost always comes from outside the configure, so it is
 	#  not something we can control.
 	# Also make makedepend aware of compiler's built-in defines.
-	if [ "$with_makedepend" != "0" ]; then
+	if [ "$with_makedepend" != "0" ] || [ "$enable_builtin_depend" != "0" ]; then
 		cflags_makedep="`echo | $cxx_host -E -x c++ -dM - | sed 's~.define ~-D~g;s~ .*~ ~g;s~(.*)~~g' | tr -d '\r\n'`"
 		cflags_makedep="$cflags_makedep `echo "$CFLAGS" | sed 's~ /~ -~g;s~-I[ ]*[^ ]*~~g'`"
 	else
@@ -1686,6 +1700,10 @@
 }
 
 check_makedepend() {
+	if [ "$enable_builtin_depend" != "0" ]; then
+		with_makedepend="0"
+	fi
+
 	if [ "$with_makedepend" = "0" ]; then
 		log 1 "checking makedepend... disabled"
 		return
@@ -2622,6 +2640,7 @@
 		s~!!BINARY_NAME!!~$binary_name~g;
 		s~!!STRGEN!!~$STRGEN~g;
 		s~!!ENDIAN_CHECK!!~$ENDIAN_CHECK~g;
+		s~!!DEPEND!!~$DEPEND~g;
 		s~!!ENDIAN_FORCE!!~$endian~g;
 		s~!!STAGE!!~$STAGE~g;
 		s~!!MAKEDEPEND!!~$makedepend~g;
@@ -2904,6 +2923,7 @@
 	echo "  --with-icu[=icu-config]        enables icu (used for right-to-left support)"
 	echo "  --with-iconv[=iconv-path]      enables iconv support"
 	echo "  --with-psp-config[=psp-config] enables psp-config support (PSP ONLY)"
+	echo "  --disable-builtin-depend       disable use of builtin deps finder"
 	echo "  --with-makedepend[=makedepend] enables makedepend support"
 	echo ""
 	echo "Some influential environment variables:"

--- a/configure
+++ b/configure
@@ -67,6 +67,7 @@
 TTD="openttd$EXE"
 STRGEN="strgen$EXE"
 ENDIAN_CHECK="endian_check$EXE"
+DEPEND="depend$EXE"
 
 if [ -z "$sort" ]; then
 	PIPE_SORT="sed s~a~a~"

new file mode 100644
--- /dev/null
+++ b/src/depend/depend.cpp
@@ -0,0 +1,941 @@
+/* $Id$ */
+
+/**
+ * @file depend/depend.cpp Custom implementation of Makedepend.
+ *
+ * We previously used makedepend, but that could not handle the amount of
+ * files we have and does not handle conditional includes in a sane manner.
+ * This caused many link problems because not enough files were recompiled.
+ * This has lead to the development of our own dependency generator. It is
+ * meant to be a substitute to the (relatively slow) dependency generation
+ * via gcc. It thus helps speeding up compilation. It will also ignore
+ * system headers making it less error prone when system headers are moved
+ * or renamed.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <ctype.h>
+#include <limits.h>
+#include <unistd.h>
+#include <map>
+#include <set>
+#include <stack>
+
+/** Simple string comparator using strcmp as implementation */
+struct StringCompare {
+	/**
+	 * Compare a to b using strcmp.
+	 * @param a string to compare.
+	 * @param b string to compare.
+	 * @return whether a is less than b.
+	 */
+	bool operator () (const char *a, const char *b) const
+	{
+		return strcmp(a, b) < 0;
+	}
+};
+/** Set of C-style strings. */
+typedef std::set<const char*, StringCompare> StringSet;
+/** Mapping of C-style string to a set of C-style strings. */
+typedef std::map<const char*, StringSet*, StringCompare> StringMap;
+/** Pair of C-style string and a set of C-style strings. */
+typedef std::pair<const char*, StringSet*> StringMapItem;
+
+/** Include directory to search in. */
+static StringSet _include_dirs;
+/** Files that have been parsed/handled with their dependencies. */
+static StringMap _files;
+/** Dependencies of headers. */
+static StringMap _headers;
+/** The current 'active' defines. */
+static StringSet _defines;
+
+/**
+ * Helper class to read a file.
+ */
+class File {
+public:
+	/**
+	 * Create the helper by opening the given file.
+	 * @param filename the file to open
+	 * @post the file is open; otherwise the application is killed.
+	 */
+	File(const char *filename) : filename(filename)
+	{
+		this->fp = fopen(filename, "r");
+		if (this->fp == NULL) {
+			fprintf(stdout, "Could not open %s for reading\n", filename);
+			exit(1);
+		}
+		this->dirname = strdup(filename);
+		char *last = strrchr(this->dirname, '/');
+		if (last != NULL) {
+			*last = '\0';
+		} else {
+			*this->dirname = '\0';
+		}
+	}
+
+	/** Free everything we have allocated. */
+	~File()
+	{
+		fclose(this->fp);
+		free(this->dirname);
+	}
+
+	/**
+	 * Get a single character from the file.
+	 * If we are reading beyond the end of the file '\0' is returned.
+	 * @return the read character.
+	 */
+	char GetChar() const
+	{
+		char c = fgetc(this->fp);
+		return (c == EOF) ? '\0' : c;
+	}
+
+	/**
+	 * Get the directory name of the file.
+	 * @return the directory name.
+	 */
+	const char *GetDirname() const
+	{
+		return this->dirname;
+	}
+
+private:
+	FILE *fp;             ///< The currently opened file.
+	char *dirname;        ///< The directory of the file.
+	const char *filename; ///< The name of the file.
+};
+
+/** A token returned by the tokenizer. */
+enum Token {
+	TOKEN_UNKNOWN,    ///< Unknown token
+	TOKEN_END,        ///< End of document
+	TOKEN_EOL,        ///< End of line
+	TOKEN_SHARP,      ///< # character, usually telling something important comes.
+	TOKEN_LOCAL,      ///< Read a local include
+	TOKEN_GLOBAL,     ///< Read a global include
+	TOKEN_IDENTIFIER, ///< Identifier within the data.
+	TOKEN_DEFINE,     ///< (#)define in code
+	TOKEN_IF,         ///< (#)if in code
+	TOKEN_IFDEF,      ///< (#)ifdef in code
+	TOKEN_IFNDEF,     ///< (#)ifndef in code
+	TOKEN_ELIF,       ///< (#)elif in code
+	TOKEN_ELSE,       ///< (#)else in code
+	TOKEN_ENDIF,      ///< (#)endif in code
+	TOKEN_UNDEF,      ///< (#)undef in code
+	TOKEN_OR,         ///< '||' within #if expression
+	TOKEN_AND,        ///< '&&' within #if expression
+	TOKEN_DEFINED,    ///< 'defined' within #if expression
+	TOKEN_OPEN,       ///< '(' within #if expression
+	TOKEN_CLOSE,      ///< ')' within #if expression
+	TOKEN_NOT,        ///< '!' within #if expression
+	TOKEN_ZERO,       ///< '0' within #if expression
+	TOKEN_INCLUDE,    ///< (#)include in code
+};
+
+/** Mapping from a C-style keyword representation to a Token. */
+typedef std::map<const char*, Token, StringCompare> KeywordList;
+
+/**
+ * Lexer of a file.
+ */
+class Lexer {
+public:
+	/**
+	 * Create the lexer and fill the keywords table.
+	 * @param file the file to read from.
+	 */
+	Lexer(const File *file) : file(file), current_char('\0'), string(NULL), token(TOKEN_UNKNOWN)
+	{
+		this->keywords["define"]  = TOKEN_DEFINE;
+		this->keywords["defined"] = TOKEN_DEFINED;
+		this->keywords["if"]      = TOKEN_IF;
+		this->keywords["ifdef"]   = TOKEN_IFDEF;
+		this->keywords["ifndef"]  = TOKEN_IFNDEF;
+		this->keywords["include"] = TOKEN_INCLUDE;
+		this->keywords["elif"]    = TOKEN_ELIF;
+		this->keywords["else"]    = TOKEN_ELSE;
+		this->keywords["endif"]   = TOKEN_ENDIF;
+		this->keywords["undef"]   = TOKEN_UNDEF;
+
+		/* Initialise currently read character. */
+		this->Next();
+
+		/* Allocate the buffer. */
+		this->buf_len = 32;
+		this->buf = (char*)malloc(sizeof(*this->buf) * this->buf_len);
+	}
+
+	/** Free everything */
+	~Lexer()
+	{
+		free(this->buf);
+	}
+
+	/**
+	 * Read the next character into 'current_char'.
+	 */
+	void Next()
+	{
+		this->current_char = this->file->GetChar();
+	}
+
+	/**
+	 * Get the current token.
+	 * @return the token.
+	 */
+	Token GetToken() const
+	{
+		return this->token;
+	}
+
+	/**
+	 * Read the currenty processed string.
+	 * @return the string, can be NULL.
+	 */
+	const char *GetString() const
+	{
+		return this->string;
+	}
+
+	/**
+	 * Perform the lexing/tokenizing of the file till we can return something
+	 * that must be parsed.
+	 */
+	void Lex()
+	{
+		for (;;) {
+			free(this->string);
+			this->string = NULL;
+			this->token  = TOKEN_UNKNOWN;
+
+			switch (this->current_char) {
+				/* '\0' means End-Of-File */
+				case '\0': this->token = TOKEN_END; return;
+
+				/* Skip some chars, as they don't do anything */
+				case '\t': this->Next(); break;
+				case '\r': this->Next(); break;
+				case ' ':  this->Next(); break;
+
+				case '\\':
+					this->Next();
+					if (this->current_char == '\n') this->Next();
+					break;
+
+				case '\n':
+					this->token = TOKEN_EOL;
+					this->Next();
+					return;
+
+				case '#':
+					this->token = TOKEN_SHARP;
+					this->Next();
+					return;
+
+				case '"':
+					this->ReadString('"', TOKEN_LOCAL);
+					this->Next();
+					return;
+
+				case '<':
+					this->ReadString('>', TOKEN_GLOBAL);
+					this->Next();
+					return;
+
+				case '&':
+					this->Next();
+					if (this->current_char == '&') {
+						this->Next();
+						this->token = TOKEN_AND;
+						return;
+					}
+					break;
+
+				case '|':
+					this->Next();
+					if (this->current_char == '|') {
+						this->Next();
+						this->token = TOKEN_OR;
+						return;
+					}
+					break;
+
+				case '(':
+					this->Next();
+					this->token = TOKEN_OPEN;
+					return;
+
+				case ')':
+					this->Next();
+					this->token = TOKEN_CLOSE;
+					return;
+
+				case '!':
+					this->Next();
+					if (this->current_char != '=') {
+						this->token = TOKEN_NOT;
+						return;
+					}
+					break;
+
+				/* Possible begin of comment */
+				case '/':
+					this->Next();
+					switch (this->current_char) {
+						case '*': {
+							this->Next();
+							char previous_char = '\0';
+							while ((this->current_char != '/' || previous_char != '*') && this->current_char != '\0') {
+								previous_char = this->current_char;
+								this->Next();
+							}
+							this->Next();
+							break;
+						}
+						case '/': while (this->current_char != '\n' && this->current_char != '\0') this->Next(); break;
+						default: break;
+					}
+					break;
+
+				default:
+					if (isalpha(this->current_char) || this->current_char == '_') {
+						/* If the name starts with a letter, it is an identifier */
+						this->ReadIdentifier();
+						return;
+					}
+					if (isdigit(this->current_char)) {
+						bool zero = this->current_char == '0';
+						this->Next();
+						if (this->current_char == 'x' || this->current_char == 'X') Next();
+						while (isdigit(this->current_char) || this->current_char == '.' || (this->current_char >= 'a' && this->current_char <= 'f') || (this->current_char >= 'A' && this->current_char <= 'F')) {
+							zero &= this->current_char == '0';
+							this->Next();
+						}
+						if (zero) this->token = TOKEN_ZERO;
+						return;
+					}
+					this->Next();
+					break;
+			}
+		}
+	}
+
+private:
+	/**
+	 * The token based on keyword with a given name.
+	 * @param name the actual keyword.
+	 * @return the token of the keyword.
+	 */
+	Token FindKeyword(const char *name) const
+	{
+		KeywordList::const_iterator it = this->keywords.find(name);
+		if (it == this->keywords.end()) return TOKEN_IDENTIFIER;
+		return (*it).second;
+	}
+
+	/**
+	 * Read an identifier.
+	 */
+	void ReadIdentifier()
+	{
+		size_t count = 0;
+
+		/* Read the rest of the identifier */
+		do {
+			this->buf[count++] = this->current_char;
+			this->Next();
+
+			if (count >= buf_len) {
+				/* Scale the buffer if required */
+				this->buf_len *= 2;
+				this->buf = (char *)realloc(this->buf, sizeof(*this->buf) * this->buf_len);
+			}
+		} while ((isalpha(this->current_char) || this->current_char == '_' || isdigit(this->current_char)));
+		this->buf[count] = '\0';
+
+		free(this->string);
+		this->string = strdup(this->buf);
+		this->token = FindKeyword(this->string);
+	}
+
+	/**
+	 * Read a string up to a given character, then set the given token.
+	 * @param end the 'marker' for the end of the string.
+	 * @param token the token to set after returning.
+	 */
+	void ReadString(char end, Token token)
+	{
+		size_t count = 0;
+		this->Next();
+		while (this->current_char != end && this->current_char != ')' && this->current_char != '\n' && this->current_char != '\0') {
+			this->buf[count++] = this->current_char;
+			this->Next();
+
+			if (count >= this->buf_len) {
+				/* Scale the buffer if required */
+				this->buf_len *= 2;
+				this->buf = (char *)realloc(this->buf, sizeof(*this->buf) * this->buf_len);
+			}
+		}
+		this->buf[count] = '\0';
+		free(this->string);
+		this->string = strdup(this->buf);
+		this->token = token;
+	}
+
+	const File *file;     ///< The file to read from.
+	char current_char;    ///< The current character to process.
+	char *string;         ///< Currently processed string.
+	Token token;          ///< The current token to process.
+	char *buf;            ///< Temporary buffer.
+	size_t buf_len;       ///< Length of the temporary buffer.
+	KeywordList keywords; ///< All keywords we know of.
+};
+
+/**
+ * Generate a path from a directory name and a relative filename.
+ * If the file is not local the include directory names will be used instead
+ * of the passed parameter with directory name. If the file is local both will
+ * be queried where the parameter takes precedence.
+ * @param dirname the directory to look in.
+ * @param filename the file to look for.
+ * @param local whether to look locally (in dirname) for the file.
+ * @return the absolute path, or NULL if the file doesn't exist.
+ */
+const char *GeneratePath(const char *dirname, const char *filename, bool local)
+{
+	if (local) {
+		if (access(filename, R_OK) == 0) return strdup(filename);
+
+		char path[PATH_MAX];
+		strcpy(path, dirname);
+		const char *p = filename;
+		/* Remove '..' from the begin of the filename. */
+		while (*p == '.') {
+			if (*(++p) == '.') {
+				*(strrchr(path, '/')) = '\0';
+				p += 2;
+			}
+		}
+		strcat(path, "/");
+		strcat(path, p);
+
+		if (access(path, R_OK) == 0) return strdup(path);
+	}
+
+	for (StringSet::iterator it = _include_dirs.begin(); it != _include_dirs.end(); it++) {
+		char path[PATH_MAX];
+		strcpy(path, *it);
+		const char *p = filename;
+		/* Remove '..' from the begin of the filename. */
+		while (*p == '.') {
+			if (*(++p) == '.') {
+				*(strrchr(path, '/')) = '\0';
+				p += 2;
+			}
+		}
+		strcat(path, "/");
+		strcat(path, p);
+
+		if (access(path, R_OK) == 0) return strdup(path);
+	}
+
+	return NULL;
+}
+
+/**
+ * Try to parse a 'defined(expr)' expression.
+ * @param lexer the lexer to get tokens from.
+ * @param defines the set of known defines.
+ * @param verbose whether to give verbose debugging information.
+ * @return the value of the expression.
+ */
+bool ExpressionDefined(Lexer *lexer, StringSet *defines, bool verbose);
+
+/**
+ * Try to parse a 'expr || expr' expression.
+ * @param lexer the lexer to get tokens from.
+ * @param defines the set of known defines.
+ * @param verbose whether to give verbose debugging information.
+ * @return the value of the expression.
+ */
+bool ExpressionOr(Lexer *lexer, StringSet *defines, bool verbose);
+
+/**
+ * Try to parse a '!expr' expression. Also parses the '(expr)', '0' and
+ * identifiers. Finally it also consumes any unknown tokens.
+ * @param lexer the lexer to get tokens from.
+ * @param defines the set of known defines.
+ * @param verbose whether to give verbose debugging information.
+ * @return the value of the expression.
+ */
+bool ExpressionNot(Lexer *lexer, StringSet *defines, bool verbose)
+{
+	if (lexer->GetToken() == TOKEN_NOT) {
+		if (verbose) fprintf(stderr, "!");
+		lexer->Lex();
+		bool value = !ExpressionDefined(lexer, defines, verbose);
+		if (verbose) fprintf(stderr, "[%d]", value);
+		return value;
+	}
+
+	if (lexer->GetToken() == TOKEN_OPEN) {
+		if (verbose) fprintf(stderr, "(");
+		lexer->Lex();
+		bool value = ExpressionOr(lexer, defines, verbose);
+		if (verbose) fprintf(stderr, ")[%d]", value);
+		lexer->Lex();
+		return value;
+	}
+
+	if (lexer->GetToken() == TOKEN_ZERO) {
+		if (verbose) fprintf(stderr, "0");
+		lexer->Lex();
+		if (verbose) fprintf(stderr, "[0]");
+		return false;
+	}
+
+	bool first = true;
+	while (lexer->GetToken() == TOKEN_UNKNOWN || lexer->GetToken() == TOKEN_IDENTIFIER) {
+		if (verbose && first) fprintf(stderr, "<assumed true>");
+		first = false;
+		lexer->Lex();
+	}
+
+	return true;
+}
+
+/**
+ * Try to parse a 'defined(expr)' expression.
+ * @param lexer the lexer to get tokens from.
+ * @param defines the set of known defines.
+ * @param verbose whether to give verbose debugging information.
+ * @return the value of the expression.
+ */
+bool ExpressionDefined(Lexer *lexer, StringSet *defines, bool verbose)
+{
+	bool value = ExpressionNot(lexer, defines, verbose);
+
+	if (lexer->GetToken() != TOKEN_DEFINED) return value;
+	lexer->Lex();
+	if (verbose) fprintf(stderr, "defined");
+	bool open = (lexer->GetToken() == TOKEN_OPEN);
+	if (open) lexer->Lex();
+	if (verbose) fprintf(stderr, open ? "(" : " ");
+	if (lexer->GetToken() == TOKEN_IDENTIFIER) {
+		if (verbose) fprintf(stderr, "%s", lexer->GetString());
+		value = defines->find(lexer->GetString()) != defines->end();
+	}
+	if (open) {
+		if (verbose) fprintf(stderr, ")");
+		lexer->Lex();
+	}
+	lexer->Lex();
+	if (verbose) fprintf(stderr, "[%d]", value);
+	return value;
+}
+
+/**
+ * Try to parse a 'expr && expr' expression.
+ * @param lexer the lexer to get tokens from.
+ * @param defines the set of known defines.
+ * @param verbose whether to give verbose debugging information.
+ * @return the value of the expression.
+ */
+bool ExpressionAnd(Lexer *lexer, StringSet *defines, bool verbose)
+{
+	bool value = ExpressionDefined(lexer, defines, verbose);
+
+	while (true) {
+		if (lexer->GetToken() != TOKEN_AND) return value;
+		if (verbose) fprintf(stderr, " && ");
+		lexer->Lex();
+		value = value && ExpressionDefined(lexer, defines, verbose);
+	}
+}
+
+/**
+ * Try to parse a 'expr || expr' expression.
+ * @param lexer the lexer to get tokens from.
+ * @param defines the set of known defines.
+ * @param verbose whether to give verbose debugging information.
+ * @return the value of the expression.
+ */
+bool ExpressionOr(Lexer *lexer, StringSet *defines, bool verbose)
+{
+	bool value = ExpressionAnd(lexer, defines, verbose);
+
+	while (true) {
+		if (lexer->GetToken() != TOKEN_OR) return value;
+		if (verbose) fprintf(stderr, " || ");
+		lexer->Lex();
+		value = value || ExpressionAnd(lexer, defines, verbose);
+	}
+}
+
+/** Enumerator to tell how long to ignore 'stuff'. */
+enum Ignore {
+	NOT_IGNORE,         ///< No ignoring.
+	IGNORE_UNTIL_ELSE,  ///< Ignore till a #else is reached.
+	IGNORE_UNTIL_ENDIF, ///< Ignore till a #endif is reached.
+};
+
+/**
+ * Scan a file for includes, defines and the lot.
+ * @param filename the name of the file to scan.
+ * @param ext the extension of the filename.
+ * @param header whether the file is a header or not.
+ * @param verbose whether to give verbose debugging information.
+ */
+void ScanFile(const char *filename, const char *ext, bool header, bool verbose)
+{
+	static StringSet defines;
+	static std::stack<Ignore> ignore;
+	/* Copy in the default defines (parameters of depend) */
+	if (!header) {
+		for (StringSet::iterator it = _defines.begin(); it != _defines.end(); it++) {
+			defines.insert(strdup(*it));
+		}
+	}
+
+	File file(filename);
+	Lexer lexer(&file);
+
+	/* Start the lexing! */
+	lexer.Lex();
+
+	while (lexer.GetToken() != TOKEN_END) {
+		switch (lexer.GetToken()) {
+			/* We reached the end of the file... yay, we're done! */
+			case TOKEN_END: break;
+
+			/* The line started with a # (minus whitespace) */
+			case TOKEN_SHARP:
+				lexer.Lex();
+				switch (lexer.GetToken()) {
+					case TOKEN_INCLUDE:
+						if (verbose) fprintf(stderr, "%s #include ", filename);
+						lexer.Lex();
+						switch (lexer.GetToken()) {
+							case TOKEN_LOCAL:
+							case TOKEN_GLOBAL: {
+								if (verbose) fprintf(stderr, "%s", lexer.GetString());
+								if (!ignore.empty() && ignore.top() != NOT_IGNORE) {
+									if (verbose) fprintf(stderr, " (ignored)");
+									break;
+								}
+								const char *h = GeneratePath(file.GetDirname(), lexer.GetString(), lexer.GetToken() == TOKEN_LOCAL);
+								if (h != NULL) {
+									StringMap::iterator it = _headers.find(h);
+									if (it == _headers.end()) {
+										it = (_headers.insert(StringMapItem(strdup(h), new StringSet()))).first;
+										if (verbose) fprintf(stderr, "\n");
+										ScanFile(h, ext, true, verbose);
+									}
+									StringMap::iterator curfile;
+									if (header) {
+										curfile = _headers.find(filename);
+									} else {
+										/* Replace the extension with the provided extension of '.o'. */
+										char path[PATH_MAX];
+										strcpy(path, filename);
+										*(strchr(path, '.')) = '\0';
+										strcat(path, ext != NULL ? ext : ".o");
+										curfile = _files.find(path);
+										if (curfile == _files.end()) {
+											curfile = (_files.insert(StringMapItem(strdup(path), new StringSet()))).first;
+										}
+									}
+									if (it != _headers.end()) {
+										for (StringSet::iterator header = it->second->begin(); header != it->second->end(); header++) {
+											if (curfile->second->find(*header) == curfile->second->end()) curfile->second->insert(strdup(*header));
+										}
+									}
+									if (curfile->second->find(h) == curfile->second->end()) curfile->second->insert(strdup(h));
+									free((void*)h);
+								}
+							}
+							/* Fall through */
+							default: break;
+						}
+						break;
+
+					case TOKEN_DEFINE:
+						if (verbose) fprintf(stderr, "%s #define ", filename);
+						lexer.Lex();
+						if (lexer.GetToken() == TOKEN_IDENTIFIER) {
+							if (verbose) fprintf(stderr, "%s", lexer.GetString());
+							if (!ignore.empty() && ignore.top() != NOT_IGNORE) {
+								if (verbose) fprintf(stderr, " (ignored)");
+								break;
+							}
+							if (defines.find(lexer.GetString()) == defines.end()) defines.insert(strdup(lexer.GetString()));
+							lexer.Lex();
+						}
+						break;
+
+					case TOKEN_UNDEF:
+						if (verbose) fprintf(stderr, "%s #undef ", filename);
+						lexer.Lex();
+						if (lexer.GetToken() == TOKEN_IDENTIFIER) {
+							if (verbose) fprintf(stderr, "%s", lexer.GetString());
+							if (!ignore.empty() && ignore.top() != NOT_IGNORE) {
+								if (verbose) fprintf(stderr, " (ignored)");
+								break;
+							}
+							StringSet::iterator it = defines.find(lexer.GetString());
+							if (it != defines.end()) {
+								free((void*)*it);
+								defines.erase(it);
+							}
+							lexer.Lex();
+						}
+						break;
+
+					case TOKEN_ENDIF:
+						if (verbose) fprintf(stderr, "%s #endif", filename);
+						lexer.Lex();
+						if (!ignore.empty()) ignore.pop();
+						if (verbose) fprintf(stderr, " -> %signore", (!ignore.empty() && ignore.top() != NOT_IGNORE) ? "" : "not ");
+						break;
+
+					case TOKEN_ELSE: {
+						if (verbose) fprintf(stderr, "%s #else", filename);
+						lexer.Lex();
+						Ignore last = ignore.empty() ? NOT_IGNORE : ignore.top();
+						if (!ignore.empty()) ignore.pop();
+						if (ignore.empty() || ignore.top() == NOT_IGNORE) {
+							ignore.push(last == IGNORE_UNTIL_ELSE ? NOT_IGNORE : IGNORE_UNTIL_ENDIF);
+						} else {
+							ignore.push(IGNORE_UNTIL_ENDIF);
+						}
+						if (verbose) fprintf(stderr, " -> %signore", (!ignore.empty() && ignore.top() != NOT_IGNORE) ? "" : "not ");
+						break;
+					}
+
+					case TOKEN_ELIF: {
+						if (verbose) fprintf(stderr, "%s #elif ", filename);
+						lexer.Lex();
+						Ignore last = ignore.empty() ? NOT_IGNORE : ignore.top();
+						if (!ignore.empty()) ignore.pop();
+						if (ignore.empty() || ignore.top() == NOT_IGNORE) {
+							bool value = ExpressionOr(&lexer, &defines, verbose);
+							ignore.push(last == IGNORE_UNTIL_ELSE ? (value ? NOT_IGNORE : IGNORE_UNTIL_ELSE) : IGNORE_UNTIL_ENDIF);
+						} else {
+							ignore.push(IGNORE_UNTIL_ENDIF);
+						}
+						if (verbose) fprintf(stderr, " -> %signore", (!ignore.empty() && ignore.top() != NOT_IGNORE) ? "" : "not ");
+						break;
+					}
+
+					case TOKEN_IF: {
+						if (verbose) fprintf(stderr, "%s #if ", filename);
+						lexer.Lex();
+						if (ignore.empty() || ignore.top() == NOT_IGNORE) {
+							bool value = ExpressionOr(&lexer, &defines, verbose);
+							ignore.push(value ? NOT_IGNORE : IGNORE_UNTIL_ELSE);
+						} else {
+							ignore.push(IGNORE_UNTIL_ENDIF);
+						}
+						if (verbose) fprintf(stderr, " -> %signore", (!ignore.empty() && ignore.top() != NOT_IGNORE) ? "" : "not ");
+						break;
+					}
+
+					case TOKEN_IFDEF:
+						if (verbose) fprintf(stderr, "%s #ifdef ", filename);
+						lexer.Lex();
+						if (lexer.GetToken() == TOKEN_IDENTIFIER) {
+							bool value = defines.find(lexer.GetString()) != defines.end();
+							if (verbose) fprintf(stderr, "%s[%d]", lexer.GetString(), value);
+							if (ignore.empty() || ignore.top() == NOT_IGNORE) {
+								ignore.push(value ? NOT_IGNORE : IGNORE_UNTIL_ELSE);
+							} else {
+								ignore.push(IGNORE_UNTIL_ENDIF);
+							}
+						}
+						if (verbose) fprintf(stderr, " -> %signore", (!ignore.empty() && ignore.top() != NOT_IGNORE) ? "" : "not ");
+						break;
+
+					case TOKEN_IFNDEF:
+						if (verbose) fprintf(stderr, "%s #ifndef ", filename);
+						lexer.Lex();
+						if (lexer.GetToken() == TOKEN_IDENTIFIER) {
+							bool value = defines.find(lexer.GetString()) != defines.end();
+							if (verbose) fprintf(stderr, "%s[%d]", lexer.GetString(), value);
+							if (ignore.empty() || ignore.top() == NOT_IGNORE) {
+								ignore.push(!value ? NOT_IGNORE : IGNORE_UNTIL_ELSE);
+							} else {
+								ignore.push(IGNORE_UNTIL_ENDIF);
+							}
+						}
+						if (verbose) fprintf(stderr, " -> %signore", (!ignore.empty() && ignore.top() != NOT_IGNORE) ? "" : "not ");
+						break;
+
+					default:
+						if (verbose) fprintf(stderr, "%s #<unknown>", filename);
+						lexer.Lex();
+						break;
+				}
+				if (verbose) fprintf(stderr, "\n");
+			/* Fall through */
+			default:
+				/* Ignore the rest of the garbage on this line */
+				while (lexer.GetToken() != TOKEN_EOL && lexer.GetToken() != TOKEN_END) lexer.Lex();
+				lexer.Lex();
+				break;
+		}
+	}
+
+	if (!header) {
+		for (StringSet::iterator it = defines.begin(); it != defines.end(); it++) {
+			free((void*)*it);
+		}
+		defines.clear();
+		while (!ignore.empty()) ignore.pop();
+	}
+}
+
+/**
+ * Entry point. Arguably the most common function in all applications.
+ * @param argc the number of arguments.
+ * @param argv the actual arguments.
+ * @return return value for the caller to tell we succeed or not.
+ */
+int main(int argc, char *argv[])
+{
+	bool ignorenext = true;
+	char *filename = NULL;
+	char *ext = NULL;
+	bool append = false;
+	bool verbose = false;
+
+	for (int i = 0; i < argc; i++) {
+		if (ignorenext) {
+			ignorenext = false;
+			continue;
+		}
+		if (argv[i][0] == '-') {
+			/* Append */
+			if (strncmp(argv[i], "-a", 2) == 0) append = true;
+			/* Include dir */
+			if (strncmp(argv[i], "-I", 2) == 0) {
+				if (argv[i][2] == '\0') {
+					i++;
+					_include_dirs.insert(strdup(argv[i]));
+				} else {
+					_include_dirs.insert(strdup(&argv[i][2]));
+				}
+				continue;
+			}
+			/* Define */
+			if (strncmp(argv[i], "-D", 2) == 0) {
+				char *p = strchr(argv[i], '=');
+				if (p != NULL) *p = '\0';
+				_defines.insert(strdup(&argv[i][2]));
+				continue;
+			}
+			/* Output file */
+			if (strncmp(argv[i], "-f", 2) == 0) {
+				if (filename != NULL) continue;
+				filename = strdup(&argv[i][2]);
+				continue;
+			}
+			/* Object file extension */
+			if (strncmp(argv[i], "-o", 2) == 0) {
+				if (ext != NULL) continue;
+				ext = strdup(&argv[i][2]);
+				continue;
+			}
+			/* Verbose */
+			if (strncmp(argv[i], "-v", 2) == 0) verbose = true;
+			continue;
+		}
+		ScanFile(argv[i], ext, false, verbose);
+	}
+
+	/* Default output file is Makefile */
+	if (filename == NULL) filename = strdup("Makefile");
+
+	char backup[PATH_MAX];
+	strcpy(backup, filename);
+	strcat(backup, ".bak");
+
+	char *content = NULL;
+	long size = 0;
+
+	/* Read in the current file; so we can overwrite everything from the
+	 * end of non-depend data marker down till the end. */
+	FILE *src = fopen(filename, "rb");
+	if (src != NULL) {
+		fseek(src, 0, SEEK_END);
+		size = ftell(src);
+		rewind(src);
+		content = (char*)malloc(size * sizeof(*content));
+		fread(content, 1, size, src);
+		fclose(src);
+	}
+
+	FILE *dst = fopen(backup, "wb");
+	if (content != NULL) fwrite(content, 1, size, dst);
+	fclose(dst);
+
+	/* Then append it to the real file. */
+	src = fopen(backup, "r");
+	dst = fopen(filename, "w");
+	while (fgets(content, size, src) != NULL) {
+		fputs(content, dst);
+		if (!append && !strncmp(content, "# DO NOT DELETE", 15)) break;
+	}
+	if (ftell(src) == 0) fprintf(dst, "\n# DO NOT DELETE\n");
+	for (StringMap::iterator it = _files.begin(); it != _files.end(); it++) {
+		for (StringSet::iterator h = it->second->begin(); h != it->second->end(); h++) {
+			fprintf(dst, "%s: %s\n", it->first, *h);
+		}
+	}
+
+	/* Clean up our mess. */
+	fclose(dst);
+	fclose(src);
+
+	free(filename);
+	free(ext);
+	free(content);
+
+	for (StringMap::iterator it = _files.begin(); it != _files.end(); it++) {
+		for (StringSet::iterator h = it->second->begin(); h != it->second->end(); h++) {
+			free((void*)*h);
+		}
+		it->second->clear();
+		delete it->second;
+		free((void*)it->first);
+	}
+	_files.clear();
+
+	for (StringMap::iterator it = _headers.begin(); it != _headers.end(); it++) {
+		for (StringSet::iterator h = it->second->begin(); h != it->second->end(); h++) {
+			free((void*)*h);
+		}
+		it->second->clear();
+		delete it->second;
+		free((void*)it->first);
+	}
+	_headers.clear();
+
+	for (StringSet::iterator it = _defines.begin(); it != _defines.end(); it++) {
+		free((void*)*it);
+	}
+	_defines.clear();
+
+	for (StringSet::iterator it = _include_dirs.begin(); it != _include_dirs.end(); it++) {
+		free((void*)*it);
+	}
+	_include_dirs.clear();
+
+	return 0;
+}