ruby-em-socksify build depends on hpsocksd which has been removed
(see #1093570)

it was initially packaged for diaspora, but not needed anymore.

Thanks!

tags 1094501 moreinfo
thanks

I see the removal request for http-request but webmock looks like it's
sticking around. Mind taking a look into these?

Checking reverse dependencies...
# Broken Depends:
ruby-em-http-request: ruby-em-http-request

# Broken Build-Depends:
ruby-em-http-request: ruby-em-socksify
ruby-webmock: ruby-em-socksify

This is also part of the current key package set.

Paul

Hi,
d-i:

meta:manual	=>	debian-installer
debian-installer	build-depends	dns323-firmware-tools
dns323-firmware-tools	depends	ruby-ffi
ruby-ffi	build-depends	ruby-rspec
ruby-rspec	build-depends	ruby-childprocess
ruby-childprocess	build-depends	ruby-coveralls
ruby-coveralls	build-depends	ruby-vcr
ruby-vcr	build-depends	ruby-em-http-request
ruby-em-http-request	build-depends	ruby-em-socksify

ISTM d-i could stop build-depending dns323-firmware-tools. There
are no armel kernels anymore (except -rpi). Keeping a firmware
support package for armel hardware that cannot be installed onto
anymore seems unnecessary to me.

Maybe there are other reasons that will keep ruby-em-socksify in the
key package set, but this chain could be cut off?

Adding CC: debian-boot@lists.debian.org so they could hopefully
chime in.

BR,
Chris

Hi,

Chris Hofstaedtler <zeha@debian.org> (2025-03-23):

We don't build d-i for armel anymore, so dns323-firmware-tools can be
removed from the list of key packages.

I haven't tried to clean things up after dropping armel and i386 since
it looked like (at least) the latter was still interesting to keep for
someone (https://bugs.debian.org/1084783#12).


Cheers,

Hi,

And I think that's not the only reason. I haven't check all the paths,
but my analysis (attached) shows a lot of things going on. Mind you, it
does look like there are some places where you don't need many cuts.

Paul

last direct reverse dependency.