looo wrote: ↑
Sat Mar 17, 2018 9:04 pm
I guess a simple statement to not support py2 for the next release /release cycle will fix this issue. I know that is painful, but it doesn't get much better if we wait until 2020 or die with py2...
My two cents:
I think we should go this way. Make into the Roadmap of v0.18 only supporting py3. The reasons are:
1. Py2 and Py3 compatibility sinks effort and developer time.
2. With the release of v0.17 it ends the introduction of py3 incompatible code. It should be a requirement that anything merged into v0.18 is py3 compatible.
3. v0.17, once released, will be short-living, as we plan to release v0.18 around the end of this year. This will allow to have two reasonably similar versions from a functionality point of view, one supporting python2 (v0.17), the other python3 (v0.18). I think this is a perfect bridge (at least when faced with the egg/chicken problem) to disrupt users as little as possible, while allowing to create awareness and move all other tools, 3rd party WB and macros that have not been adapted during v0.18 development cycle (hopefully very little of them), during v0.19 development cycle. This should allow to bring all python tools together for the release of v0.19, in time for the 2020 shutdown.
Probably 3rd party python tools will need either to adapt to have two branches, one for py2 (addonmanager of v0.17 pointing there) and another for py3( addonmanager of v0.18 pointing there). Hopefully the development of such tools for py3 will gain moment from the decision of the Roadmap for v0.18.