Vitis AI new upstream version

[Łukasz Zemczak]

[Impact]

Xilinx has released the 2.0 version of Vitis AI. To make the latest tooling/development-stack available to all Ubuntu Xilinx users, it is important for us to get the latest version available for the officially supported Xilinx releases (jammy and focal).

[Test Case]

The update will be mainly tested by Xilinx and the HWE team.

More on testing shortly.

[Regression Potential]

As this is a new upstream version, it needs to be noted that there is a lot of risk of regressing existing users. Upstream has not yet added autopkgtests, but there are plans to get that done. In the meantime however, Xilinx already did some regression testing using VAI 2.0 built from source on the Ubuntu 20.04 ZCU102 target with the V1.3.2 DPU IP in the Programable Logic and reported no issues.

It is worth noting that the *only* consumers of the package are users of Ubuntu on Xilinx platforms: https://ubuntu.com/download/xilinx . There is no risk of regressing any other machines as these packages are not usable there.

[Launchpad Janitor (janitor)]

This bug was fixed in the package vitis-ai - 2.0-0ubuntu1

---------------
vitis-ai (2.0-0ubuntu1) jammy; urgency=medium

  [ Łukasz 'sil2100' Zemczak ]
  * New upstream release (LP: #1959382).
    - Exclude multiple invalid files from the upstream tarball that would
      make this package incompatible with universe.
  * debian/patches: drop unneeded patches, refresh still relevant ones.
  * Add debian/patches/fix-python-install.patch to allow installing the python
    modules to the debian/tmp directory for package builds.
  * Add debian/patches/fix-ftbfs-includes-and-namespaces.patch to fix
    compliation errors due to missing includes or namespaces.
  * Drop vitis-ai-dnndk as it is no longer present since 1.4+.
  * Install all the newly added Vitis-AI libraries.
  * Tweak debian/copyright to be more accurate.
  * Build python libraries and utilities for Vitis-AI-Library.

  [ Alfonso Sanchez-Beato ]
  * Move to the newer name for xrt package (xrt-embedded -> xrt)
  * Make sure all VART libraries are in vitis-ai-runtime package
  * Replace fix for -Werror=nonnull with something maybe safer

 -- Łukasz 'sil2100' Zemczak <email address hidden> Fri, 28 Jan 2022 12:01:04 +0100