[oe] [meta-java][RFC][PATCH 7/7] ca-certificates-java: switch to more generic / more correct JAVA (R)DEPENDS

[Kyle Russell]

On Fri, Jul 27, 2018 at 9:29 AM André Draszik <git at andred.net> wrote:

> From: André Draszik <andre.draszik at jci.com>
>
> This recipe here actually depends on Java7 upwards, not Java2 upwards,
> so state that dependency correctly using the newly introduced (virtual)
> RPROVIDES of the OpenJDK8/OpenJRE8 recipes.
>
> At the same time, we can remove hard-coded references to openjdk-8 and
> openjre-8, and thusly make the build more flexible.
>
> Signed-off-by: André Draszik <andre.draszik at jci.com>
> ---
>  .../ca-certificates-java_20180516.bb                   | 10 ++--------
>  1 file changed, 2 insertions(+), 8 deletions(-)
>
> diff --git a/recipes-core/ca-certificates-java/
> ca-certificates-java_20180516.bb b/recipes-core/ca-certificates-java/
> ca-certificates-java_20180516.bb
> index a2d7abd..2ae7352 100644
> --- a/recipes-core/ca-certificates-java/ca-certificates-java_20180516.bb
> +++ b/recipes-core/ca-certificates-java/ca-certificates-java_20180516.bb
> @@ -14,7 +14,7 @@ DEPENDS = "virtual/javac-native fastjar-native"
>

Should this DEPEND on virtual/java7-sdk-native instead?

 # We can't use virtual/javac-native, because that would create a
>  # keystore that can't be read on the target (as virtual/javac-native
>  # usually is either too old, or plain incompatible with this)
> -PACKAGE_WRITE_DEPS += "openjdk-8-native"
> +PACKAGE_WRITE_DEPS += "virtual/java7-runtime-native"
>

Would you still need the PACKAGE_WRITE_DEPS if the DEPENDS above was
changed?  I thought virtual/javac-native was "too old" because it was
reserved for the bootstrap process, so I've been considering it as a
special case.