add-on depends library updates

[AlwinEsch]

v21.3.5

• Update depednencies to fix building on apple silicon (gnutls 3.8.4) and binutils on linux (2.41 or higher)

To have newest versions on dependencies and to fix problem on #284

[AlwinEsch]

Should now ready for review

[phunkyfish]

Nice, thanks!

[AlwinEsch]

Should it be becoming a version increase and release or has you planned additional changes on add-on about?

[fuzzard]

Hold off on this. There are serious questions regarding the upstream xz-utils tarballs. See team repo for details

[fuzzard]

Drop the xz-utils bump for now thanks.

[AlwinEsch]

Drop the xz-utils bump for now thanks.

Thanks @fuzzard, dropped the xzutils away.

[phunkyfish]

Strange, even though we dropped the xz-utils commit builds are still failing on Windows.

-- Using src='https://tukaani.org/xz/xz-5.4.3.tar.xz'
-- [download 100% complete]
-- [download 6% complete]
-- [download 11% complete]
-- [download 13% complete]
-- [download 31% complete]
-- [download 41% complete]
-- [download 46% complete]
-- [download 60% complete]
-- [download 83% complete]
-- [download 100% complete]
-- verifying file...
       file='C:/jenkins-workspace/workspace/binary-addons/kodi-windows-i686-Omega/cmake/addons/build/download/xz-5.4.3.tar.xz'
-- SHA256 hash of
    C:/jenkins-workspace/workspace/binary-addons/kodi-windows-i686-Omega/cmake/addons/build/download/xz-5.4.3.tar.xz
  does not match expected value
    expected: '92177bef62c3824b4badc524f8abcce54a20b7dbcfb84cde0a2eb8b49159518c'
      actual: '3881a256b0e563fc18fc1c47d8c533cdb3a311bac2920f726d01cb2449f00c97'
-- Hash mismatch, removing...

[phunkyfish]

Ah, maybe it's because the repo has been disabled:

So we can't download it the unaffected tarball.

[AlwinEsch]

Strange, even though we dropped the xz-utils commit builds are still failing on Windows.

Ah, maybe it's because the repo has been disabled:
So we can't download it the unaffected tarball.

No this was already on sunday and monday where I started on DevCon with depends update, the strange was/is, if I build locally it has always worked with sha256, but within CI's it seems to come different one.

This wrong checksum was always equal. In retrospect, this shasum error makes me very thoughtful, not that something is actually being injected with a modified package using evil proxies.

[AlwinEsch]

@phunkyfish have pushed a update to use xz-utils from our mirror, is then 5.4.1 instead of 5.4.3, but as github and his https://tukaani.org/xz no more available should older better.

[AlwinEsch]

Thanks a lot @phunkyfish for the update to version 5.4.3 and the release.

[phunkyfish]

All the thanks to you @AlwinEsch, I only did the way bits 😉