Re: [gentoo-dev] [PATCH 2/3] ruby-ng.eclass: Use eapply for RUBY_PATCHES in EAPI 6

[Hans de Graaff]

On Sun, 2017-02-19 at 09:26 +0100, Michał Górny wrote:
> W dniu 19.02.2017, nie o godzinie 09∶03 +0100, użytkownik Hans de
> Graaff
> napisał:
> >
> > This also removes the need for inheriting eutils in EAPI 6.
> > 
> Wouldn't EAPI 6 be a good opportunity to kill this dualism and just
> require people to say FILESDIR explicitly?

Which made me think: why have RUBY_PATCHES at all. I think we added
that at the time to allow bash arrays as input. So I've changed this
patch to use the default phase in EAPI 6 instead, removing the need for
explicit eapply_user and PATCHES support.

Hans

1b2b8abde99a59e72fbcc0bc8f9e835dcaee2176
Author:     Hans de Graaff <graaff@g.o>
AuthorDate: Sun Feb 19 08:42:31 2017 +0100
Commit:     Hans de Graaff <graaff@g.o>
CommitDate: Tue Feb 21 20:18:17 2017 +0100

Parent:     1d36503 ruby-ng.eclass: add support for EAPI=6
Merged:     eapi6 master
Containing: eapi6

ruby-ng.eclass: use the default src_prepare in EAPI 6

This removes the need for inheriting eutils in EAPI 6.  It also use
the standard PATCHES support instead of RUBY_PATCHES, which was
introduced to handle arrays on patches at the time.

The default phase also handles eapply_user to handle user patches.

1 file changed, 30 insertions(+), 13 deletions(-)
eclass/ruby-ng.eclass | 43 ++++++++++++++++++++++++++++++-------------

modified   eclass/ruby-ng.eclass
@@ -73,7 +73,14 @@
 # (e.g. selenium's firefox driver extension). When set this argument
is
 # passed to "grep -E" to remove reporting of these shared objects.
 
-inherit eutils java-utils-2 multilib toolchain-funcs ruby-utils
+local inherits=""
+case ${EAPI} in
+	2|3|4|5)
+		inherits="eutils"
+		;;
+esac
+
+inherit ${inherits} java-utils-2 multilib toolchain-funcs ruby-utils
 
 EXPORT_FUNCTIONS src_unpack src_prepare src_configure src_compile
src_test src_install pkg_setup
 
@@ -399,15 +406,24 @@ ruby-ng_src_unpack() {
 }
 
 _ruby_apply_patches() {
-	for patch in "${RUBY_PATCHES[@]}"; do
-		if [ -f "${patch}" ]; then
-			epatch "${patch}"
-		elif [ -f "${FILESDIR}/${patch}" ]; then
-			epatch "${FILESDIR}/${patch}"
-		else
-			die "Cannot find patch ${patch}"
-		fi
-	done
+	case ${EAPI} in
+		2|3|4|5)
+			for patch in "${RUBY_PATCHES[@]}"; do
+				if [ -f "${patch}" ]; then
+					epatch "${patch}"
+				elif [ -f "${FILESDIR}/${patch}" ];
then
+					epatch "${FILESDIR}/${patch}"
+				else
+					die "Cannot find patch
${patch}"
+				fi
+			done
+			;;
+		6)
+			if [[ -n ${RUBY_PATCHES[@]} ]]; then
+			   eqawarn "RUBY_PATCHES is no longer
supported, use PATCHES instead"
+			fi
+			;;
+	esac
 
 	# This is a special case: instead of executing just in the
special
 	# "all" environment, this will actually copy the effects on
_all_
@@ -432,14 +448,15 @@ ruby-ng_src_prepare() {
 	# almost every other ebuild.
 	find . -name '._*' -delete
 
-	_ruby_invoke_environment all _ruby_apply_patches
-
+	# Handle PATCHES and user supplied patches via the default
phase
 	case ${EAPI} in
 		6)
-			eapply_user
+			_ruby_invoke_environment all default
 			;;
 	esac
 
+	_ruby_invoke_environment all _ruby_apply_patches
+
 	_PHASE="source copy" \
 		_ruby_each_implementation _ruby_source_copy