Given the python-2 countdown deadline being 2020-01-01, a month away,
shall we get rid of python-2?
yes please as much as possible.
In my experience most server instances can live without py2 at all (after clang 9).
In desktop world thare are very few, but important packages that still require it.
The Gimp being one prominent example
If the answer is yes, we will need to decide on the following
python-2-only packages.
,----
| $ comm -23 <(equery -qC h python_targets_python2_7 | sort ) <(equery -qC h python_targets_python3_6 | sort)
| dev-lang/yasm-1.3.0
| dev-libs/libxslt-1.1.33-r1
| dev-python/backports-functools-lru-cache-1.5
| dev-python/enum34-1.1.6-r1
| dev-python/functools32-3.2.3
| dev-python/futures-3.2.0
| dev-python/pygobject-2.28.6-r55
| dev-python/pygtk-2.24.0-r4
| dev-python/subprocess32-3.2.7
| dev-util/boost-build-1.70.0
| dev-vcs/subversion-1.12.2
| gnome-base/libglade-2.6.4-r2
| net-analyzer/nmap-7.70
| sys-devel/clang-8.0.1
| x11-wm/xpra-2.4.3
`----
"As much as possible" IMHO exclude at least yasm, libxslt and nmap
If the answer is no, to avoid holding back new versions having only
python3, such as bug 671796 for dev-python/matplotlib bump, old versions
with python_targets_python2_7 and new versions without should be
co-installable into different SLOTs.
whatever your (gentoo dev) choice I wish you good luck