Thank you for doing this deep level investigation. I think this pretty
much shows that listing all modules in a way that can be maintained
isn't feasible (the idea was that maybe there's only a few of them
needed for ptest).
Which brings us back to the question: why is perl-modules using
RRECOMMEN
Following your suggestions and iteratively adding packages based on error
message, I am finally able to come up with quite a long list of perl-module-*
packages that fix the dependencies of openssl-ptest.
However, I am worried about if the list I came up with is truly comprehensive
in the sense
How are you making the list? I think this might be somewhat sped up,
if you first grep
${WORKDIR}/packages-split/openssl-ptest/usr/lib/openssl/ptest/test/
(and util/perl/OpenSSL/ as well) for "^use " and transform the output
lines into 'perl-module-something'. Then try it on the target in qemu
to s
I have iteratively added the following 45 modules, and I still not at the point
of covering all perl-module-* dependencies for openssl-test. Should I continue
on with this effort? Using "perl-module-force-all" is a lot more convenient
without conflicting with any existing usage of "perl-modules"
If this is only for fixing OpenSSL ptest, then it’s not needed perhaps,
once OpenSSL recipe is fixed?
Alex
On Tue 16. Jan 2024 at 17.56, Lyu, William via lists.openembedded.org
wrote:
> From: William Lyu
>
> When perl-modules is added via COMPLEMENTARY_GLOB, all perl-module-*
> packages recomm
From: William Lyu
When perl-modules is added via COMPLEMENTARY_GLOB, all perl-module-*
packages recommended by perl-modules are ignored due to the defined
behavior of COMPLEMENTARY_GLOB.
This patch creates a new package perl-modules-force-all that RDEPENDS on
the same set of packages perl-module
