[gentoo-dev] [PATCH v2 1/3] ruby-ng.eclass: improve error when no valid Ruby in USE_RUBY

[gentoo-dev] [PATCH v2 1/3] ruby-ng.eclass: improve error when no valid Ruby in USE_RUBY

[Sam James]

This means we don't get confusing *DEPEND/REQUIRED_USE errors about it being
unparseable and instead just get a straightforward die message indicating
the problem.

Signed-off-by: Sam James <sam@gentoo.org>
---
 eclass/ruby-ng.eclass | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/eclass/ruby-ng.eclass b/eclass/ruby-ng.eclass
index c273a431c5b1..b81038237a6b 100644
--- a/eclass/ruby-ng.eclass
+++ b/eclass/ruby-ng.eclass
@@ -103,16 +103,21 @@ ruby_implementation_depend() {
 # Return a list of valid implementations in USE_RUBY, skipping the old
 # implementations that are no longer supported.
 _ruby_get_all_impls() {
-	local i
+	local i found_valid_impl
 	for i in ${USE_RUBY}; do
 		case ${i} in
 			# removed implementations
 			ruby19|ruby20|ruby21|ruby22|ruby23|ruby24|ruby25|ruby26|jruby)
 				;;
 			*)
+				found_valid_impl=1
 				echo ${i};;
 		esac
 	done
+
+	if [[ -z ${found_valid_impl} ]] ; then
+		die "No supported implementation in USE_RUBY."
+	fi
 }
 
 # @FUNCTION: ruby_samelib
-- 
2.40.0

[gentoo-dev] [PATCH v2 2/3] ruby-ng.eclass: don't quote IUSE in has test

[Sam James]

I was looking at the generated metadata for nokogiri and noticed several
instances of 'test' in IUSE & RESTRICT. Ended up finding that the quoting
of 'IUSE' in has is wrong here.

This isn't explicitly wrong, but it's a bit ugly in the generated metadata,
and I'm not convinced Portage handles duplicates correctly (see e.g.
the duplicate || ( ... ) issue in *DEPEND).

This doesn't fix all instances and to be honest, it's a bit dubious given has_iuse
doesn't work in global scope, but let's just fix the thing which is definitely
wrong (the quoting).

This does change the metadata generated in the way I expected, but it doesn't
remove *all* duplicates.

Signed-off-by: Sam James <sam@gentoo.org>
---
 eclass/ruby-ng.eclass | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/eclass/ruby-ng.eclass b/eclass/ruby-ng.eclass
index b81038237a6b..b63dd19e18f2 100644
--- a/eclass/ruby-ng.eclass
+++ b/eclass/ruby-ng.eclass
@@ -234,7 +234,7 @@ ruby_add_rdepend() {
 		6) DEPEND="${DEPEND} test? ( ${dependency} )" ;;
 		*) BDEPEND="${BDEPEND} test? ( ${dependency} )" ;;
 	esac
-	if ! has test "$IUSE"; then
+	if ! has test ${IUSE}; then
 		IUSE+=" test"
 		RESTRICT+=" !test? ( test )"
 	fi
-- 
2.40.0

[gentoo-dev] [PATCH v2 3/3] ruby-fakegem.eclass: don't double-add USE=doc/test

[Sam James]

This isn't explicitly wrong, but it's a bit ugly in the generated metadata,
and I'm not convinced Portage handles duplicates correctly (see e.g.
the duplicate || ( ... ) issue in *DEPEND).

This doesn't fix all instances and to be honest, it's a bit dubious given has_iuse
doesn't work in global scope, but let's just fix the thing which is definitely
wrong (the quoting).

This does change the metadata generated in the way I expected, but it doesn't
remove *all* duplicates.

Signed-off-by: Sam James <sam@gentoo.org>
---
 eclass/ruby-fakegem.eclass | 42 ++++++++++++++++++++++++++------------
 1 file changed, 29 insertions(+), 13 deletions(-)

diff --git a/eclass/ruby-fakegem.eclass b/eclass/ruby-fakegem.eclass
index b3262dc5cdb4..8d22c4e3ccf7 100644
--- a/eclass/ruby-fakegem.eclass
+++ b/eclass/ruby-fakegem.eclass
@@ -150,17 +150,23 @@ RUBY_FAKEGEM_SUFFIX="${RUBY_FAKEGEM_SUFFIX:-}"
 
 case ${RUBY_FAKEGEM_RECIPE_DOC} in
 	rake)
-		IUSE+=" doc"
+		if ! has doc ${IUSE} ; then
+			IUSE+=" doc"
+		fi
 		ruby_add_bdepend "doc? ( dev-ruby/rake )"
 		RUBY_FAKEGEM_DOCDIR="doc"
 		;;
 	rdoc)
-		IUSE+=" doc"
+		if ! has doc ${IUSE} ; then
+			IUSE+=" doc"
+		fi
 		ruby_add_bdepend "doc? ( dev-ruby/rdoc )"
 		RUBY_FAKEGEM_DOCDIR="doc"
 		;;
 	yard)
-		IUSE+="doc"
+		if ! has doc ${IUSE} ; then
+			IUSE+="doc"
+		fi
 		ruby_add_bdepend "doc? ( dev-ruby/yard )"
 		RUBY_FAKEGEM_DOCDIR="doc"
 		;;
@@ -173,30 +179,40 @@ esac
 
 case ${RUBY_FAKEGEM_RECIPE_TEST} in
 	rake)
-		IUSE+=" test"
-		RESTRICT+=" !test? ( test )"
+		if ! has test ${IUSE} ; then
+			IUSE+=" test"
+			RESTRICT+=" !test? ( test )"
+		fi
 		ruby_add_bdepend "test? ( dev-ruby/rake )"
 		;;
 	rspec)
-		IUSE+=" test"
-		RESTRICT+=" !test? ( test )"
+		if ! has test ${IUSE} ; then
+			IUSE+=" test"
+			RESTRICT+=" !test? ( test )"
+		fi
 		# Also require a new enough rspec-core version that installs the
 		# rspec-2 wrapper.
 		ruby_add_bdepend "test? ( dev-ruby/rspec:2 >=dev-ruby/rspec-core-2.14.8-r2 )"
 		;;
 	rspec3)
-		IUSE+=" test"
-		RESTRICT+=" !test? ( test )"
+		if ! has test ${IUSE} ; then
+			IUSE+=" test"
+			RESTRICT+=" !test? ( test )"
+		fi
 		ruby_add_bdepend "test? ( dev-ruby/rspec:3 )"
 		;;
 	cucumber)
-		IUSE+=" test"
-		RESTRICT+=" !test? ( test )"
+                if ! has test ${IUSE} ; then
+			IUSE+=" test"
+			RESTRICT+=" !test? ( test )"
+                fi
 		ruby_add_bdepend "test? ( dev-util/cucumber )"
 		;;
 	sus)
-		IUSE+=" test"
-		RESTRICT+=" !test? ( test )"
+		if ! has test ${IUSE} ; then
+			IUSE+=" test"
+			RESTRICT+=" !test? ( test )"
+		fi
 		ruby_add_bdepend "test? ( dev-ruby/sus )"
 		;;
 	none)
-- 
2.40.0

Re: [gentoo-dev] [PATCH v2 2/3] ruby-ng.eclass: don't quote IUSE in has test

[Ulrich Mueller]

>>>>> On Wed, 29 Mar 2023, Sam James wrote:

> -	if ! has test "$IUSE"; then
> +	if ! has test ${IUSE}; then

You cannot reliably test for a flag in IUSE with code like this.
PMS defines the function in_iuse() for this (unless the above is
in global scope, in which case you're out of luck).

Ulrich

Re: [gentoo-dev] [PATCH v2 2/3] ruby-ng.eclass: don't quote IUSE in has test

[Sam James]

[-- Attachment #1: Type: text/plain, Size: 471 bytes --]


Ulrich Mueller <ulm@gentoo.org> writes:

>>>>>> On Wed, 29 Mar 2023, Sam James wrote:
>
>> -	if ! has test "$IUSE"; then
>> +	if ! has test ${IUSE}; then
>
> You cannot reliably test for a flag in IUSE with code like this.
> PMS defines the function in_iuse() for this (unless the above is
> in global scope, in which case you're out of luck).

Yep, it is, unfortunately. I can try ripping it out entirely given it's
not reliable, depending on how others feel about it.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 377 bytes --]