@michael-mycroft , @gez-mycroft had promised Mycroft AI would put out a Dinkum development roadmap by the end of January, but we’re still waiting. This is on top of the months of silence we’ve had to put up with, including the non-announcement about the very existence of Dinkum in the first place and the subpar set of skills (that doesn’t match the advertised set of skills advertised for years now) and even a lack an installation skill!
IDEA: Why not arrange something with the developer(s) who created Dinkum in the first place. Sort of like what Mycroft AI is doing with Neon? And support a community effort to get the Dinkum skills up to par, such as @clintonthegeek has called for and @jmillerv has begun.
Unfortunately this previous comment from Michael is the extent of the Dinkum roadmap for the foreseeable future:
Currently, our primary goal for Dinkum is to make it functional without depending on Selene. This is in testing now. We’ve also fixed a number of bugs. Whether we have the ability to provide further updates depends on… well, on whether we have any employees.
We’ve made the changes to ensure it remains functional even if our backend services need to be shut down. Those changes will be pushed to the Beta channel shortly for broader testing, before going out to the Stable (default) channel next week.
IDEA: Why not arrange something with the developer(s) who created Dinkum in the first place. Sort of like what Mycroft AI is doing with Neon? And support a community effort to get the Dinkum skills up to par, such as @clintonthegeek has called for and @jmillerv has begun.
Dinkum was created by our internal dev team, so that would be us. Although you’re probably referring more specifically to Mike Hansen who did a bunch of the work on it. Mike is now working at Home Assistant and has renewed his work on Rhasspy, so I am very doubtful that he’ll be contributing to future Dinkum updates.
How much attention Dinkum gets will likely depend on whether we have any employees, what partnerships we have with other organizations, and who’s managing those repositories. Even supporting a community effort takes time and someone to actually do that coordination, code review, testing and deployment. These are the less exciting but very essential parts of software development that most people don’t want to do.
Currently there is no MSM support in Dinkum, so the options would be to include any additionally desired Skills by default, create a mechanism for installation, or improve documentation to make it easier for people to load new Skill files.
@michael-mycroft@gez-mycroft Without the full set of skills that were advertised and no owner-friendly way to install new skills, the Mark II / Dinkum can not be called a consumer-ready device, or even match what you advertised it as for years and years. Mycroft AI will have to publicize a roadmap to consumer-readiness now, or its about to be hit with what could be its final deathblow, when the wider FLOSS and voice assistant worlds beyond the immediate community fully realizes the wide gap between what was promised and what was delivered and how poorly Mycroft AI has treated its own community. Surely it would be better to attempt something and maybe not succeed than to throw up your hands and self-destruct this way.
And btw, what about spinning the internal dev team off into a Dinkum / Dinkum Skills / Selene / Grokotron group so that even if Mycroft AI itself goes under, the project(s) would continue?