Merge branch 'bugfix/windows' into 'master'

Fix some Windows build issues

Fix and/or document some issues seen under Windows (especially if git clones with CRLF line endings)

See merge request !87

docs/build_system.rst

@@ -193,6 +193,9 @@ Because components usually live under the project directory (although
 they can also reside in an other folder), the path to this may be
 something like  /home/myuser/projects/myprojects/components/httpd .
 
+Components can have any name (unique to the project) but the name
+cannot contain spaces (esp-idf does not support spaces in paths).
+
 One of the things that most components will have is a component.mk makefile,
 containing instructions on how to build the component. Because the
 build environment tries to set reasonable defaults that will work most

docs/linux-setup.rst

@@ -108,6 +108,7 @@ Note the ``--recursive`` option! If you have already cloned ESP-IDF without this
     cd ~/esp/esp-idf
     git submodule update --init
 
+**IMPORTANT:** The esp-idf build system does not support spaces in paths to esp-idf or to projects.
 
 Step 3: Starting a project
 ==========================

docs/macos-setup.rst

@@ -122,6 +122,7 @@ The easiest way to start a project is to download the template project from GitH
 
 This will download ``esp-idf-template`` project into ``~/esp/myapp`` directory.
 
+**IMPORTANT:** The esp-idf build system does not support spaces in paths to esp-idf or to projects.
 
 Step 4: Building and flashing the application
 =============================================

docs/windows-setup.rst

@@ -50,6 +50,8 @@ Change to the directory you want to clone the SDK into by typing a command like
 
 If you'd rather use a Windows UI tool to manage your git repositories, this is also possible. A wide range are available.
 
+*NOTE*: While cloning submodules, the ``git clone`` command may print some output starting ``': not a valid identifier...``. This is a `known issue`_ but the git clone still succeeds without any problems.
+
 Step 3: Starting a project
 ==========================
 
@@ -59,6 +61,8 @@ The easiest way to start a project is to download the Getting Started project fr
 
 The process is the same as for checking out the ESP-IDF from github. Change to the parent directory and run ``git clone https://github.com/espressif/esp-idf-template.git``.
 
+**IMPORTANT:** The esp-idf build system does not support spaces in paths to esp-idf or to projects.
+
 Step 4: Configuring the project
 ===============================
 
@@ -74,3 +78,4 @@ If you'd like to use the Eclipse IDE instead of running ``make``, check out the
 .. _Eclipse: eclipse-setup.rst
 .. _MSYS2: https://msys2.github.io/
 .. _github: https://github.com/espressif/esp-idf-template
+.. _known issue: https://github.com/espressif/esp-idf/issues/11

make/test_build_system.sh

@@ -80,6 +80,22 @@ function run_tests()
 		 failure "Files weren't cleaned: ${ALL_BUILD_FILES}"
 	fi
 
+	print_status "Can still clean build if all text files are CRLFs"
+	make clean
+	find . -exec unix2dos {} \; # CRLFify template dir
+	# make a copy of esp-idf and CRLFify it
+	CRLF_ESPIDF=${TESTDIR}/esp-idf-crlf
+	mkdir -p ${CRLF_ESPIDF}
+	cp -rv ${IDF_PATH}/* ${CRLF_ESPIDF}
+	# don't CRLFify executable files, as Linux will fail to execute them
+	find ${CRLF_ESPIDF} -type f ! -perm 755 -exec unix2dos {} \;
+	make IDF_PATH=${CRLF_ESPIDF}
+	# do the same checks we do for the clean build
+	assert_built ${APP_BINS} ${BOOTLOADER_BINS} partitions_singleapp.bin
+	[ -f ${BUILD}/partition*.bin ] || failure "A partition table should have been built in CRLF mode"
+
+	# NOTE: If adding new tests, add them above this CRLF test...
+
 	print_status "All tests completed"
 	if [ -n "${FAILURES}" ]; then
 		echo "Some failures were detected:"

tools/kconfig/Makefile

@@ -301,7 +301,8 @@ zconf.lex.c: zconf.l
 	flex -L -P zconf -o zconf.lex.c zconf.l
 
 zconf.hash.c: zconf.gperf
-	gperf -t --output-file zconf.hash.c -a -C -E -g -k '1,3,$$' -p -t zconf.gperf
+# strip CRs on Windows systems where gperf will otherwise barf on them
+	sed -E "s/\r//" zconf.gperf | gperf -t --output-file zconf.hash.c -a -C -E -g -k '1,3,$$' -p -t
 
 zconf.tab.c: zconf.y
 	bison -t -l -p zconf -o zconf.tab.c zconf.y

tools/kconfig/zconf.l

@@ -114,8 +114,8 @@ n	[A-Za-z0-9_-]
 		zconflval.string = text;
 		return T_WORD;
 	}
-	.	warn_ignored_character(*yytext);
-	\n	{
+	[^\r\n]  warn_ignored_character(*yytext);
+	\r?\n	{
 		BEGIN(INITIAL);
 		current_file->lineno++;
 		return T_EOL;
@@ -139,7 +139,7 @@ n	[A-Za-z0-9_-]
 		new_string();
 		BEGIN(STRING);
 	}
-	\n	BEGIN(INITIAL); current_file->lineno++; return T_EOL;
+	\r?\n	BEGIN(INITIAL); current_file->lineno++; return T_EOL;
 	({n}|[/.])+	{
 		const struct kconf_id *id = kconf_id_lookup(yytext, yyleng);
 		if (id && id->flags & TF_PARAM) {
@@ -184,7 +184,7 @@ n	[A-Za-z0-9_-]
 		} else
 			append_string(yytext, 1);
 	}
-	\n	{
+	\r?\n	{
 		printf("%s:%d:warning: multi-line strings not supported\n", zconf_curname(), zconf_lineno());
 		current_file->lineno++;
 		BEGIN(INITIAL);
@@ -218,16 +218,16 @@ n	[A-Za-z0-9_-]
 			append_string("        ", ts);
 		}
 	}
-	[ \t]*\n/[^ \t\n] {
+	[ \t]*\r?\n/[^ \t\r\n] {
 		current_file->lineno++;
 		zconf_endhelp();
 		return T_HELPTEXT;
 	}
-	[ \t]*\n	{
+	[ \t]*\r?\n	{
 		current_file->lineno++;
 		append_string("\n", 1);
 	}
-	[^ \t\n].* {
+	[^ \t\r?\n].* {
 		while (yyleng) {
 			if ((yytext[yyleng-1] != ' ') && (yytext[yyleng-1] != '\t'))
 				break;